JP2010035237A - Position detection device and method of target image and program for controlling position detection device of target image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a position detection device, or the like, capable of more accurately deciding image region of a face. <P>SOLUTION: An image of a subject is captured, thus obtaining a subject image α1 of the N-1th frame and a subject image α2 of the N-th frame. Face image detection processing is performed in the subject images α1, α2 of the respective frames, thus detecting face image regions C1, C2, C3. The face image region C2 has been linked, since there is the corresponding face image region C1. The face image region C3 has been linked, since a corresponding face image region does not exist. For the subject image α2 of the N-th frame, the linked face image region C2 is decided as a face image region. The face image region can be decided relatively accurately. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a target image position detection apparatus and method, and a program for controlling the target image position detection apparatus.

Background of the Invention

  When the subject is a person or the like, it is often preferable that the image portion of the face of the subject is captured clearly. For this reason, the image portion of the face may be detected from the subject image. In order to improve the accuracy of face detection, a plurality of faces may be detected for the same image (Patent Document 1).

  Also, when a continuous subject image is given, a region near the face image portion detected in the previous frame subject image is defined as a region for detecting the face image portion in the current frame subject image. Yes (Patent Document 2).

Japanese Patent Laid-Open No. 2003-216935 Japanese Patent Laid-Open No. 11-15979

  However, erroneous detection in the subject image of the previous frame may affect detection in the subject image of the subsequent frame. Detection accuracy may be lowered.

  An object of the present invention is to improve the detection accuracy of an object such as a face.

  The target image position detection apparatus according to the present invention includes a first target image from a first subject image among a plurality of subject images obtained by simultaneously or successively capturing subjects that can be regarded as substantially the same. First detection means for detecting a position, second detection means for detecting a second position where the target image is present from a second subject image among the plurality of subject images, and the first detection means. Based on the detected first position and the second position detected by the second detecting means, the target image in at least one of the first subject image and the second subject image. It is characterized by comprising a determining means for determining the position where the sigma exists.

  The present invention also provides a method suitable for the position detection apparatus for the target image. That is, this method detects a first position where a target image exists from a first subject image among a plurality of subject images obtained by simultaneously or successively capturing subjects that can be regarded as substantially the same, A second position where the target image is present is detected from a second subject image among the plurality of subject images, and the first position is determined based on the detected first position and the detected second position. The position where the target image exists in at least one of the subject image and the second subject image is determined.

  The present invention also provides a program for realizing the above-described target image position detection method. Furthermore, a recording medium storing the program is also provided.

  According to the present invention, a plurality of subject images obtained by imaging subjects that can be regarded as substantially the same simultaneously or successively (subject images obtained continuously include, for example, pre-imaging, video imaging, continuous shooting, etc. , The first position where the target image exists is detected from the first subject image of the subject image obtained in one imaging sequence such as bracket imaging. A second position where the target image exists is also detected from the second subject image among the subject images of the plurality of frames.

  Based on the detected first position and second position, a position where the target image exists in at least one of the first subject image and the second subject image is determined. Since the position where the target image exists is determined by comparing the first position and the second position, the first position detected in the first subject image or the second subject image is detected. The detection accuracy is improved as compared with the case where the position of the target image is detected based on only one of the second positions. A detection process with relatively high accuracy can be realized.

  There may be further provided an image pickup apparatus that picks up images of subjects that can be regarded as substantially the same or successively and outputs image data representing a plurality of subject images. In this case, the first detection means detects a position where the target image exists from the first subject image among the plurality of subject images represented by the image data output from the imaging device. The second detection means will detect a position where the target image exists from the second subject image among the plurality of subject images represented by the image data output from the imaging device.

  The first detection means and the second detection means may be common.

  A position where the size of the first subject image is different from the size of the second subject image, and the size of the first subject image is the same as the size of the second subject image; As described above, it may further comprise position conversion means for converting at least one of the first position detected by the first detection means and the second position detected by the second detection means. . In this case, the determining means will determine the position where the target image exists based on the first position and the second position converted by the position converting means.

  Even if the size of the first subject image is different from the size of the second subject image, the position where the target image exists can be determined based on the first position and the second position.

  An imaging device that images a subject and outputs image data representing the subject image, and the imaging device that images the subject and outputs first image data representing the first subject image in response to a pre-imaging command And a second imaging control unit that controls the imaging apparatus to image a subject and output second image data representing the second subject image in response to a main imaging command. You may further provide an imaging control means. In this case, the first detection means includes a first subject image represented by the first image data output from the first imaging device under the control of the first imaging control means. From which the target image is present, and the second detection means outputs second image data output from the second imaging device under the control of the second imaging control means. The position where the target image exists is detected from the second subject image represented by

  In the case of a digital camera or the like in which main imaging is performed in response to the shutter release button being turned on and image data obtained by the main imaging is recorded on a recording medium such as a memory card, the main imaging is performed. The position where the target image exists from the first position obtained from the first subject image obtained by the imaging (pre-imaging) and the second position obtained from the second subject image obtained by the main imaging. Can be detected.

  Determination means for determining whether or not imaging according to the main imaging command is performed in synchronization with strobe light emission, and the determination means when the determination means determines that the imaging is performed in synchronization with strobe light emission. It is preferable to further comprise a determination control means for controlling the determination means so as to determine the second position detected by the second detection means and the position where the target image exists, by stopping the determination processing in .

  When the main imaging is performed in synchronization with the strobe light emission, the first subject image obtained by the pre-imaging may be dark. On the other hand, since the second subject image obtained by imaging in synchronization with the flash emission has a relatively appropriate brightness, the target image is detected relatively accurately. There are many cases. However, the target image may not be detected because the first subject image is dark. Since the position where the target image exists may not be determined even by comparing the first position obtained from the first subject image and the second position obtained from the second subject image, the determination process is stopped. The The second position obtained from the second subject image obtained in synchronization with the flash emission is determined as the position where the target image exists. False detection can be prevented beforehand.

  You may further provide the exposure amount adjustment means which adjusts the exposure amount of the said imaging device based on the 1st image data output from the said imaging device based on the control in a said 1st imaging control means. In this case, the control in the second imaging control unit causes the second image data to be output from the imaging device in response to the imaging command based on the exposure amount adjusted by the exposure amount adjustment unit. Let's go. And according to the determination means for determining whether or not the exposure amount of the image pickup apparatus controlled by the first image pickup control means is less than or equal to the threshold value, and the determination means determines that the exposure value is less than or equal to the threshold value A decision control means for controlling the decision means so as to decide the second position detected by the second detection means as the position where the target image exists, by stopping the decision process in the decision means; Prepare.

  Even when it is determined that the exposure amount is equal to or less than the threshold value, the reliability of the first position is low as in the processing for the strobe light emission. For this reason, the second position is determined as the position where the target image exists.

  For example, the determining means is configured when the amount of variation between the first position detected by the first detecting means and the second position detected by the second detecting means is equal to or greater than a predetermined threshold value. , It is determined that the target image does not exist in at least one of the first subject image and the second subject image. If the fluctuation amount is equal to or greater than the threshold value, it is considered that there is a different target image or false detection, and therefore it is determined that there is no target image.

  The first detection means includes a first subject image obtained by first imaging among a plurality of subject images obtained by imaging a subject that can be regarded as substantially the same multiple times in succession in time. Alternatively, the first position where the target image exists may be detected. In this case, the second detection means detects the second position where the target image exists from the second subject image obtained by the second imaging among the subject images of the plurality of frames. The apparatus further includes third detection means for detecting a third position where the target image is present from a third subject image obtained by third imaging among the subject images of the plurality of frames. Further, the determination means includes a first position detected by the first detection means, a second position detected by the second detection means, and a third position detected by the third detection means. Based on the position, the position where the target image exists in at least one of the first subject image, the second subject image, and the third subject image will be determined.

  Of the first subject image, the second subject image, and the third subject image, when the target image cannot be detected in the second subject image for some reason, the first subject image and the second subject image The target image can be detected using the three subject images of the third subject image. Even when the target image exists, the case where the target image cannot be found can be avoided.

  The determining means includes a third position in which there is no third position detected by the third detecting means corresponding to the second position detected by the second detecting means; Corresponding to both the second position detected by the second detection means and the third position detected by the third detection means among the first positions detected by the first detection means The position where the target image exists in the third subject image may be determined based on the first position where there is no such object.

  A first imaging device that outputs first image data representing the first subject image by imaging the subject, and second image data representing the second subject image by imaging the subject. You may further provide the 2nd imaging device to do. In this case, the first detection means will detect the position where the target image exists from the first subject image represented by the first image data output from the first imaging device. Further, the second detection means will detect a position where the target image exists from the second subject image represented by the second image data output from the second imaging device.

  The first imaging device and the second imaging device may be built in the same housing or may be built in different housings.

  The first imaging device and the second imaging device may have the same or different imaging ranges.

  The first detection means detects, for example, a first position where the target image exists from the first subject image based on the value of the target image likelihood. The second detection means also detects, for example, the second position where the target image exists from the second subject image based on the target image likelihood value. In this case, the determining means is based on the target image likelihood value of the target image detected by the first detection means and the target image likelihood value of the target image detected by the second detection means, The position where the target image exists will be determined.

  A color filter of a plurality of colors is regularly provided on the light receiving surface, and further includes a single-plate solid-state electronic imaging device that outputs a plurality of color image data representing a color subject image by imaging the subject. Also good. In this case, the first subject image is represented by one color image data among the plurality of color image data output from the single-plate solid-state electronic imaging device, and the second subject image is It is represented by color image data of a color different from the color image data of one color among the color image data of a plurality of colors.

  It may further include a solid-state electronic imaging device that includes a plurality of solid-state electronic imaging elements provided with color filters of different colors on the light-receiving surface, and outputs color image data of different color types by imaging the subject. . In this case, the first subject image is represented by one type of color image data among the color image data output from the solid-state electronic imaging device, and the second subject image is the color image data. Of these, the one type of color image data will be represented by a different type of color image data.

  A visible light sensor that outputs visible light image data that represents the first subject image as a visible light image by capturing the subject, and a red that represents the second subject image as an infrared light image by capturing the subject. An infrared light sensor that outputs external light image data may be further provided. In this case, the first detection means detects a position where the target image exists from a first subject image represented by visible light image data output from the visible light sensor, and the second detection means. The detecting means detects a position where the target image exists from a second subject image among a plurality of subject images represented by infrared light image data output from the infrared light sensor. .

  On the light receiving surface, a visible light filter (a filter having a characteristic of transmitting visible light) and an infrared light filter (a filter having a characteristic of transmitting infrared light) are included. A visible light / infrared light solid-state electronic imaging device that outputs visible light image data that represents a subject image as a visible light image and that outputs infrared light image data that represents a second subject image as an infrared light image. You may prepare. In this case, the first detection means detects a position where the target image exists from the first subject image represented by the visible light image data output from the visible light / infrared light solid-state electronic imaging device. The second detection means detects a position where the target image exists from the second subject image represented by the infrared light image data output from the visible light / infrared light solid-state electronic imaging device. It will be a thing.

  There may be further provided an imaging device that images the subject at a fixed period and alternately outputs the first image data representing the subject image of the first field and the second image data representing the subject image of the second field. . In this case, the first detection means detects a position where the target image exists from the subject image of the first field represented by the image data of the first field output from the imaging device, The second detection means will detect the position where the target image exists from the second subject image represented by the image data of the second field output from the imaging device.

  The imaging device, for example, captures subjects that can be regarded as being substantially the same, changes the imaging conditions, and simultaneously or successively images them, and outputs image data representing a plurality of frames of subject images.

  The change in the imaging condition is, for example, at least one of exposure amount, aperture value, shutter speed, focus amount, zoom amount, sensitivity, and flash change.

  You may further provide the signal processing means which image | photographs a to-be-photographed object and performs a different signal process to the image data showing the object image of one frame. In this case, the first detection unit is a position where the target image exists from the first subject image represented by the image data on which one of the signal processing performed by the signal processing unit is performed. May be detected. In this case, the second detection means is based on the second subject image represented by the image data subjected to another signal processing different from the one signal processing among the signal processing performed by the signal processing means. The position where the target image exists will be detected. The first position and the second position can be detected from the first subject image and the second subject image obtained by performing different signal processing.

  The determining means is a position in the vicinity of the first position where there is a first position detected by the first detecting means corresponding to the second position detected by the second detecting means. May be determined as a position where the target image exists in the subject image.

  Hardware control means for performing hardware control based on the location of the target image determined by the determination means may be further provided.

  The hardware control is, for example, at least one of automatic exposure control, automatic focusing control, flash control, power saving control, and zoom amount control.

  You may further provide the signal processing means which performs a signal processing based on the presence position of the said target image determined in the said determination means. Predetermined signal processing can be performed on the target image existing at the detected position.

  The signal processing means performs at least one signal processing of, for example, gain adjustment, white balance adjustment, color correction, contour enhancement, noise reduction, and gradation correction.

  The target image is, for example, a face or eye image.

It is a block diagram which shows the electric constitution of a digital still camera. It is an example of a subject image. (N-1) The face detection result of the subject image of the frame is shown. The subject image face detection result of the Nth frame is shown. It is an example of a collation result. It is a flowchart which shows a face image detection process procedure. A through image, a real captured image, and the like are shown. It is a flowchart which shows a face image detection process procedure. It is a flowchart which shows a face image detection process procedure. (A)-(D) are an example of the fluctuation | variation of a face image area | region. It is a flowchart which shows the process sequence of linking. It is an example of a subject image. It is a flowchart which shows a face image detection process procedure. It is a flowchart which shows a face image detection process procedure. The relationship between a subject and an imaging device is shown. It is an example of a subject image. It is a flowchart which shows a face image detection process procedure. It is a flowchart which shows a face image detection process procedure. (A) and (B) are examples of subject images. (A) to (C) are examples of subject images. (A) and (B) are examples of a composite image. It is an example of a synthesized image. It is an example of the light-receiving surface of a single-plate image sensor. (A)-(C) are examples of a monochromatic subject image. (A)-(C) are examples of a monochromatic subject image that has been interpolated. (A)-(C) have shown the light-receiving surface of the three-plate solid-state electronic image sensor. (A) shows the light receiving surface of the visible light sensor, and (B) shows the light receiving surface of the infrared sensor. The light-receiving surface of a visible light / infrared sensor is shown. It is a time chart which shows the reading process of a field image, and a face image detection processing procedure. It is a time chart which shows the reading process of a field image, and a face image detection processing procedure. It is a time chart which shows the reading process of a field image, and a face image detection processing procedure. It is a flowchart which shows the detection process procedure of a face image area | region. It is an example of a to-be-photographed image. It is an example of a to-be-photographed image. It is an example of a to-be-photographed image. It is a flowchart which shows a face image area | region detection processing procedure. It is a flowchart which shows an automatic exposure control process. It is a flowchart which shows an automatic exposure control process. It is a flowchart which shows an automatic focusing control process. It is a flowchart which shows an automatic focusing control process. It is a flowchart which shows a gradation correction process. It is a flowchart which shows a gradation correction process. It is a flowchart which shows a color correction process procedure. It is a flowchart which shows a color correction process procedure.

  FIG. 1 shows an embodiment of the present invention and is a block diagram showing an electrical configuration of a digital still camera.

  The overall operation of the digital still camera is controlled by the control circuit 2. The control circuit 2 includes a CPU 3, a ROM 4 in which an operation program for performing an operation described later is stored, and a RAM 5 in which data and the like are temporarily stored. It is also possible to provide a digital still camera with a communication function and receive the transmitted operation program to perform the operation described later. Of course, if the operation program can be stored in the memory card, the operation program can be read from the memory card and installed in the digital still camera.

  The digital still camera includes operation buttons 1 such as a shutter release button and a mode setting switch. An operation signal output from the operation button 1 is input to the control circuit 2. The digital still camera is provided with a strobe light emitting device 7 controlled by a strobe control circuit 6.

  When the imaging mode is set by the mode setting switch, the light beam is condensed by the imaging lens 8 controlled by the lens driving circuit 11, and the light receiving surface of the imaging element 10 is passed through the diaphragm 9 controlled by the diaphragm driving circuit 12. Incident on top. A subject image is formed on the light receiving surface of the image sensor 10. The image sensor 10 is controlled by the image sensor control circuit 13, and a video signal representing a subject image (a so-called through image signal indicating a through image) is input to the analog signal processing circuit 14 at a constant imaging cycle (pre-imaging). In the analog signal processing circuit 14, predetermined analog signal processing such as correlated double sampling is performed. The video signal output from the analog signal processing circuit 14 is converted into digital image data in the analog / digital signal processing circuit and input to the digital signal processing circuit 16. In the digital signal processing circuit 16, predetermined signal processing such as gamma correction and white balance adjustment is performed. The image data output from the digital signal processing circuit 16 is given to the display device 19 via the memory 18. The captured subject image is displayed on the display screen of the display device 19.

  When the shutter release button is pressed, the image data output from the digital signal processing circuit 16 as described above is given to the memory 18 and temporarily stored. Image data is output from the memory 18 and recorded on the memory card 20.

  In the digital still camera according to this embodiment, a face image area can be detected from a subject image. For this purpose, a face detection processing circuit 17 is provided in the digital still camera. The image data output from the digital signal processing circuit 16 as described above is also input to the face detection processing circuit 17. In the face detection processing circuit 17, a region indicating a face image portion is detected as described later.

  In this embodiment, at least two frames of the subject image are used. A face image portion detection process is performed on the subject image of each frame. The area of the face image portion is determined based on the area of the face image portion detected in the subject image of each frame.

  FIG. 2 is an example of a subject image for explaining the face image area detection processing according to this embodiment.

  As described above, when the imaging mode is set, the subject is repeatedly imaged at a constant period, and the (N-1) th frame subject image α1 and the Nth frame subject image α2 are obtained. A process of detecting a face image in each of the two frames of the subject images α1 and α2 is performed. For example, the subject image α1 of the (N−1) th frame includes the person image S1, and the face image area C1 is detected by performing the face image detection process. The N-th subject image α2 includes a person image S2, and a face image area C2 is detected. The face image area C2 is considered to be at a position corresponding to the face image area C1 detected in the subject image α1 of the (N-1) th frame. In this embodiment, the area C2 is the area C1. Stipulated as being linked to. Further, in the subject image α2 of the Nth frame, the face image area C3 is also detected in the face detection process. The face image area C3 is the area where the corresponding area is not detected in the subject image α1 of the (N−1) th frame, and is not linked.

  In this embodiment, a collation process is performed between the detection result of the face image in the (N-1) th frame subject image α1 and the detection result of the face image in the Nth frame subject image α2. The detection area of the face image in the N-th frame subject image α2 is a linked area. Therefore, in the subject image α2 of the Nth frame, the unlinked area C3 is not set as the face image area, and the linked area C2 is determined as the face image area.

  In this embodiment, the face image area is also detected from the top and bottom. In FIG. 2, in the areas C1 to C3 of the detected face image, the upper part of the face image is shown thicker than the other parts.

  FIG. 3 shows the face detection result of the subject image α1 of the (N−1) th frame, and FIG. 4 shows the face detection result of the subject image α2 of the Nth frame.

  In face detection processing, the number of detected face image areas, the face number indicating the number of detected face image areas, the position of the face image area, the size of the face image area, and the in-plane rotation of the face image area A corner, an out-of-plane rotation angle of the face image area, and a score are detected. The number of face image areas and the like are stored as face detection results corresponding to the subject image. The in-plane rotation angle obtained as a face detection result is the tilt angle of the face when the face is facing the front. The out-of-plane rotation angle is an angle when the face faces sideways with respect to the case where the face faces the front.

  FIG. 5 shows the collation result.

  As described above, the detection result of the face image of the subject image α1 of the (N−1) th frame is collated with the detection result of the face image of the subject image α2 of the Nth frame. As a result of the collation, the face number, position, and size corresponding to the linked area as described above are stored in the collation result table.

  FIG. 6 is a flowchart showing the procedure for determining the face image area.

  When the imaging mode is set as described above, the subject is imaged at a constant period, and a specific one-frame subject image is obtained (step 31). Face detection processing is performed on the obtained one-frame subject image (step 32), and the face detection result is registered (step 33).

  If the previous (previous frame) face detection result is not registered (NO in step 34), the processing from step 31 is repeated.

  If the previous face detection result is registered (YES in step 34), as described above, the collation process between the previous face detection result and the current face detection result is performed (step 35). Based on the result of the collation process, a collation result representing the face image detection area to be linked as described above is registered (step 36). The face image area specified by the information registered in the collation result is determined as the face image area of the current subject image.

  The processing from step 31 is repeated until there is a termination command from the user (step 37).

  When the face image area corresponding to the face image area detected in the current face image detection process has been detected in the previous face image detection process, it is determined as the face image area in the current subject image. The face image area is not determined unless both the current and previous detections are detected. Since the area in which the face image area is erroneously detected from the current subject image is excluded from the face image area, the detection accuracy of the face image area is improved. Of course, if a face image area corresponding to the face image area detected in the previous face image detection process is also detected in the current face image detection process, the face image detection area is detected as the face image detection in the previous subject image. Needless to say, it may be determined as a region.

  7 and 8 show another embodiment.

  In this embodiment, a collation process of a detection result between a so-called through image (image at the time of pre-imaging) and a main captured image (image captured after the shutter release button is fully pressed) is performed. In addition, the captured image has a high resolution and a large size (data amount, image size), whereas a through image has a relatively low resolution and a small size. For this reason, in this embodiment, coordinate conversion is performed, and coordinate conversion processing is performed so that the size of the through image is the same as the size of the actual captured image. The through image having the same size and the actual captured image are collated with the detection result.

  Referring to FIG. 7, a through image α4 is obtained. The through image α4 includes a person image S4, and a face image region C4 is detected by performing face image detection processing. Further, the actual imaging is performed, and the actual captured image α6 is obtained. Since the size of the through image α4 is different from the size of the main captured image α6, the through image α4 is coordinate-transformed so that the through image α4 has the same size as the main captured image α6. By converting the through image α4, a through image α5 having the same size as the actual captured image α6 is obtained.

  The person image S5 is also included in the coordinate-converted through image α5, and the face image region C5 is detected by performing the face image detection process. The captured image α6 also includes the person image S6, and the face image area C6 is detected by performing the face image detection process. It is also assumed that a face image area C7 other than the face image area C6 has been detected. It can be understood that the face image region in the main captured image α6 is determined as the region C6 by collating the detection result in the through image α5 subjected to coordinate conversion and the detection result in the main captured image α6.

  In the above example, the size of the through image α4 is adjusted to the size of the main captured image α6. However, it is needless to say that the size of the main captured image α6 may be adjusted to the size of the through image α4. .

  FIG. 8 is a flowchart showing a processing procedure for determining a face image area from the through image and the main captured image.

  When the imaging mode is set and a through image is obtained by pre-imaging, face detection processing of the through image is performed (step 41). Since the through image is obtained at a constant cycle, face detection processing is performed for each through image obtained at the constant cycle. The face detection result obtained in the face detection process is registered while being updated (step 42). Until the shutter release button is fully pressed (step 43), the processing of steps 41 and 42 is repeated.

  When the shutter release button is fully pressed (YES in step 43), the main imaging is performed and a main captured image is obtained. Face detection processing is performed on the obtained captured image (step 44).

  When the actual captured image and the through image (through image immediately before the shutter release button is fully pressed) are obtained, coordinate conversion is performed so that the sizes of the images are the same (step 45). This coordinate transformation may be performed to fit the size of the through image, may be made to fit the size of the actual captured image, or may be performed to fit the size of another image. . When the coordinate conversion is performed, the face detection result of the through image and the face detection result of the main captured image are collated (step 46). The collation result is registered and the face image area is determined (step 47).

  FIG. 9 shows a modification of FIG. 8, and is a flowchart showing the processing procedure of the face image area determination processing. In this figure, the same processes as those shown in FIG.

  In the processing shown in FIG. 9, it is confirmed whether the exposure amount of the obtained through image is small or whether the strobe light is emitted in the main imaging (step 51). When strobe light is emitted in the main imaging, the exposure amount of the obtained through image is small. When the through image exposure amount is small, the face image region may not be detected even if face detection processing is performed because the through image is dark. Since the face image area cannot be detected, as described above, there is no face image area to be linked even if the face image area detected in the through image and the face image area detected in the main imaging area are collated. Sometimes. For this reason, the face image area may not be determined. For this reason, in this modification, when the exposure amount of the through image is small or the strobe light is emitted in the main imaging (YES in step 51), the above-described detection result comparison between the through image and the main imaging image is performed. The face detection result in the main captured image is registered without being processed, and is determined as a face detection area (step 52).

  FIGS. 10A to 10D and FIG. 11 are for explaining the above-described processing for determining the linking.

  FIGS. 10A to 10D show fluctuations between the face image area detected in the previous subject image and the face image region detected in the current subject image.

  (A) shows the position variation. This indicates that the face image area C8 was detected in the face detection process for the previous subject image, and the face image area C9 was detected in the face detection process for the current subject image.

  (B) shows a magnitude | size fluctuation | variation. This shows that the face image area C10 was detected in the face detection process for the previous subject image, and the face image area C11 was detected in the face detection process for the current subject image. The size of the face image area C11 is larger than the size of the face image area C10.

  (C) shows in-plane rotational fluctuation. This shows that the face image area C12 was detected in the face detection process for the previous subject image, and the face image area C13 was detected in the face detection process for the current subject image. As described above, the area corresponding to the upper part of the face is shown thick. Therefore, it can be seen that the face image detected in the current subject image is rotated 90 degrees in the plane of the subject image from the face image detected in the previous subject image.

  (D) shows the out-of-plane rotation fluctuation. This indicates that the face image area C14 was detected in the face detection process for the previous subject image, and the face image area C15 was detected in the face detection process for the current subject image. The face image area C14 detected in the face detection process in the previous subject image is a square, and the face image area C15 detected in the face detection process in the current subject image is a pentagon. In the previous subject image, it can be seen that the face image faces the front, and in the current subject image, the face image faces sideways.

  When such a change occurs, linking is performed if the degree of change is within a certain range as described below. In addition, when linking is performed, an area in the vicinity of the detected face image area or the like can be determined as the face image area. For example, as shown in (A), the area C surrounding the face image areas C8 and C9 may be determined as the face image area, or the face image areas C8 and C9 may be determined as the face image area.

  FIG. 11 is a flowchart showing a processing procedure for determining linking. When the above-described variation is within a certain range, the face image area detected in the face detection process in the previous subject image is linked with the face image area detected in the face detection process in the current subject image as a correspondence. Is done.

  The face number of the face image area detected in the current subject image is a variable j, and the face number of the face image area detected in the previous subject image is a variable i.

  All of the face image areas detected in the current subject image are initialized (steps 61 to 64). For this reason, the variable j of the current face image area is reset (step 61), and the link initialization process is repeated while the variable j is incremented until the links for all the face image areas are initialized (step 61). In step 62, YES, 63, 64).

  When all the face image areas detected in the current subject image are initialized (NO in step 62), the variable j is reset again (step 65), and one of the face image areas detected in the current subject image is reset. Linking is determined for each one and each of the face image areas detected in the previous subject image. It is confirmed whether or not linking has been completed for all of the face image areas detected in the current subject image (step 66).

  If not completed (YES in step 66), a dummy initial value is set to the above-described position variation distance (step 67).

  The variable i of the face image area detected in the previous subject image is reset (step 68). Since all face image areas are linked, it is determined whether or not all face image areas have been linked (step 69).

  If linking has not been completed for all face image regions (YES in step 69), it is confirmed whether the out-of-plane rotation variation, the in-plane rotation variation, and the size variation are within a predetermined range (steps 70 to 72). ). If all of these fluctuations are within a predetermined range (YES in steps 70 to 72), it is considered that the fluctuation amount is small. The variable d is set to the center-to-center distance in the position variation (step 73), and it is confirmed whether the center-to-center distance is smaller than the dummy value (step 74). If it is smaller (YES in step 74), the minimum value of the center-to-center distance is set as the center-to-center distance in the position variation (step 75). If the set center-to-center distance is within a predetermined range (YES in step 76), the face image area j in the current subject image as a determination target is linked to the face image area i in the previous subject image. Is registered as the above-described collation result (see FIG. 5). The variable i is incremented (step 79), and it is determined whether or not the next face image area of the previous subject image is linked.

  FIG. 12 shows another embodiment and is an example of a subject image.

  In this embodiment, a continuous three-frame subject image is used. Of three consecutive subject images, the face image region corresponding to the face image region detected in the third subject image is present in the first subject image even if the face image region does not exist in the second subject image Are linked and determined as a face image area in the third subject image.

  For example, a subject image α7 of the (N-2) th frame, a subject image α8 of the (N-1) th frame, and a subject image α9 of the Nth frame are obtained. The subject image α7 of the (N-2) th frame includes person images S7 and S8, and the face image areas C16 and C17 are detected by the face image detection process. The subject image α8 of the (N−1) th frame includes person images S9 and S10. It is assumed that the face image area C18 is detected from the person image S9 by the face image detection process, but the face image area is not detected from the person image S10. The subject image α9 of the Nth frame includes person images S11 and S12, and the face image areas C20 and C21 are detected by the face image detection process.

  Processing for collating detection results between the face image areas C20 and C21 detected in the subject image α9 of the Nth frame and the face image area C18 detected in the subject image α8 of the (N−1) th frame (first detection) Result collation) is performed. In the subject image α8 of the (N−1) th frame, there is a face image region C18 corresponding to the face image region C20 detected in the subject image α9 of the Nth frame, but the subject image of the Nth frame. A face image area corresponding to the face image area C21 detected at α9 is not detected. For this reason, in the above-described embodiment, the face image area C20 is determined as the face image area in the N-th subject image α9, but the face image area C21 is not determined as the face image area. In this embodiment, among the face image areas detected in the N-th subject image α9, the unlinked face image area C21 is linked to the face image area of the subject image α7 of the (N-2) th frame. It is confirmed whether there is something attached (second detection result collation). When the face image area C21 linked to the face image area C17 of the subject image α7 of the (N-2) th frame is in the subject image α9 of the Nth frame, the face image area C21 is It is determined as a face image area in the N-th frame subject image α9. In the above embodiment, the face image area C21 that could not be detected can be detected.

  In the subject image α8 of the (N−1) th frame, the face image region C18 is obtained by performing a detection result matching process with the face image regions C16 and C17 detected in the subject image α7 of the Nth frame. It will be understood that the face image area is determined.

  13 and 14 are flowcharts showing the processing procedure of FIG. 12 described above.

  The subject image of the current time (Nth frame) is read (step 31), and face detection processing is performed (step 32). The result of the current face detection process is registered (step 33). If the previous (N-1th frame) face detection result is registered (YES in step 34), as described above, the previous and current face detection result matching processing (first detection result matching) ) Is performed (step 52).

  In the first detection result collation, it is confirmed whether or not there is an unlinked face image area detected in the current subject image (step 53). If there is an unlinked one, it is confirmed whether the detection result of the previous time ((N-2) th frame) is registered (step 54). If the previous detection result is registered (YES in step 54), “the current detection result that is not linked to the previous detection result” and “the previous detection result“ current, previous, previous time ” The detection result collating process is performed with the “no link” (second detection result collation) (step 55).

  The first detection result matching and the second detection result matching are combined (step 56), the matching result of the combination is reflected and registered as the final detection result, and the face image in the current subject image A region is determined (step 36). The processing from step 31 is repeated until an end command is issued.

  If there is no link in the current detection result (NO in step 53), or if the previous detection result is not registered (NO in step 54), the current detection result will be based on the comparison process of the previous and current detection results. The face image area in the subject image is determined (step 36).

  FIG. 15 shows another embodiment, and shows how the subject 150 is imaged.

  In this embodiment, two digital still cameras 151 and 152 having the same imaging range are used. Two frames of the subject image are obtained by imaging the subject 150 simultaneously or successively using the two digital still cameras 151 and 152. The first digital still cameras 151 and 152 are connected to the computer device 153. Image data representing a subject image captured by the digital still cameras 151 and 153 is given to the computer device 153. In the computer device 153, the face image area is determined using the obtained two frames of the subject image. In FIG. 15, two digital still cameras 151 and 152 are shown. However, it is sufficient if there are two image pickup devices (elements), and two image pickup devices are built in one camera. Good.

  FIG. 16 shows an example of a subject image taken using the digital still cameras 151 and 152 shown in FIG.

  The digital still cameras 151 and 152 have the same imaging range, and simultaneously image a subject to obtain a first subject image α11 and a second subject image α12. The first subject image α11 includes a person image S13. By performing face detection processing on the first subject image α11, a face image region C22 is detected from the person image S13. Further, the face image area C23 is also detected in the first subject image α11. The second subject image α12 includes a person image S14. By performing face detection processing on the second subject image α12, a face image region C24 is detected from the person image S14. Further, a face image area C25 is also detected in the second subject image α12.

  The face image regions C22 and C24 are ring-attached by collating the detection results of the face image region detected in the first subject image α11 and the face image region detected in the second subject image α12. Will be. Therefore, the face image area in the first subject image α11 (may be the second subject image α12) is determined as the face image region C22 of the person image S13.

  FIG. 17 is a flowchart showing a processing procedure for determining a face image area using the first subject image α11 and the second subject image α12.

  As described above, the first subject image and the second subject image are obtained by simultaneously imaging the subject by the first digital still camera 151 and the second digital still camera 152. These first subject image and second subject image are read (steps 61 and 63). The face image detection process (step 62) in the read first subject image and the face image detection process (step 64) in the second subject image are performed in parallel.

  A collation process is performed between the result of the face image detection process in the first subject image and the result of the face image detection process in the second subject image (step 65). The collation result is reflected and registered as a detection result (step 66).

  FIG. 18 is a flowchart showing a modification of the processing procedure of FIG. In this figure, the same processes as those shown in FIG. 17 are denoted by the same reference numerals.

  A first subject image and a second subject image are obtained (steps 61 and 63). In the process shown in FIG. 17, the face image detection process in the first subject image and the face image detection process in the second subject image are performed in parallel. A face image detection process in the first subject image is performed (step 62), and then a face image detection process in the second subject image is performed (step 64). Thereafter, the collation process of the detection result (step 65) and the detection result are registered (step 66).

  FIGS. 19A and 19B to FIGS. 22A and 22B show other modified examples and are examples of subject images.

  19 (A) and 19 (B) show the subject at the same time using the first digital still camera 151 and the second digital still camera 152 whose imaging ranges partially match each other but do not completely match. The subject image obtained by imaging is shown.

  The subject image α13 shown in (A) includes person images S15 and S16. In the face detection process, the score of the face portion of the person image S15 is 8, and the score of the face portion of the person image S16 is 4. The subject image α14 shown in (B) includes person images S17 and S18. In the face detection process, the score of the face portion of the person image S17 is 4, and the score of the face portion of the person image S18 is 6. The person image S16 and the person image S17 are images of the same person.

  20 (A) to 20 (C) show how to determine the face image area when the score is obtained as described above in FIGS. 19 (A) and 19 (B).

  In this manner, the face image area is determined by individually determining the score detected in each of the subject images α13 and α14. Referring to (A) and (B), the face image portion of the human image S15 having a score of 8 and the face image portion of the human image S18 having a score of 6 each have a high score (for example, determined as a face image). The score threshold is determined to be score 5) face image areas C25 and C26. Since the human images S16 and S17 having the score 4 have a low score, they are not recognized as face image areas.

  In this way, the same portions of the subject images α13 and α14 in which the face image area is detected are superimposed (coordinate conversion), and a composite subject image α17 shown in (C) is obtained. Needless to say, the positional relationship between the digital still cameras 151 and 152 is set in the computer device 153 for coordinate conversion.

  FIGS. 21A and 21B show composite subject images α18 and α19 in which the subject images α13 and α14 are superimposed on the same portion.

  Referring to (A), the score of the face image portion of the person image S25 is 8, and the score of the face image portion of the person image S16 is the object image α13 as shown in FIGS. The score 8 is obtained by adding the score 4 of the person image S16 and the score 4 of the person image S17 in the subject image α14.

  Since the face image portion of the person image S16 has a score of 8, the face image area C28 is determined as shown in FIG.

  FIG. 22 shows composite subject images α20 and α21 in which subject images α13 and α14 are superimposed on the same part.

  The face image area is determined after weighting is performed using the weighting coefficient. A score obtained by face detection is set as a temporary score, and a score obtained by multiplying the temporary score by a weighting coefficient is set as a score. The weighting coefficient of the face image portion of the person image S15 is 1, its provisional score is 8, and the score is 8. As shown in FIGS. 20A and 20B, the score of the face image portion of the person image S16 is obtained by multiplying the provisional score 4 of the person image S16 in the subject image α13 by the weighting coefficient 1 and the person in the subject image α14. The score 12 is obtained by adding the provisional score 4 of the image S17 to the value obtained by constantly twisting the weighting coefficient 2. The weighting coefficient of the face image portion of the person image S18 is 2, its provisional score is 8, and the score is 16.

  When the threshold value of the score determined as the face image area is 7.5, all the scores are equal to or higher than the score. Therefore, as shown in FIG. 21B, each face image of the person images S15, S16, and S18. The portions are determined as face image areas C25, C28 and C26.

  FIG. 23 is an example of a light receiving surface of a single-plate image sensor.

  On the light-receiving surface of the single-plate image sensor, an R filter (indicated by R) that transmits a red light component, a G filter (indicated by G) that transmits a green light component, and a B filter that transmits a blue light component (Indicated by B) are regularly arranged at a constant period.

  A red signal component is obtained by photoelectrically converting the light component that has passed through the R filter. With the obtained red signal component, a subject image of the red component is obtained as shown in FIG. Similarly, a green signal component is obtained by photoelectrically converting the light component transmitted through the G filter. With the obtained green signal component, a subject image of the green component is obtained as shown in FIG. A blue signal component is obtained by photoelectrically converting the light component transmitted through the B filter. With the obtained blue signal component, a subject image of the blue component is obtained as shown in FIG.

  As shown in FIG. 16, face image detection processing is performed using two subject images of the subject image of the red signal component, the subject image of the green signal component, and the subject image of the blue signal component thus obtained. Thus, the face image area can be determined.

  25A, 25B, and 25C show the subject image of the red signal component, the subject image of the green signal component, and the subject image of the blue signal component after the interpolation processing. As described above, the face image area may be determined by performing face image detection processing as shown in FIG. 16 using two of the three subject images subjected to interpolation processing. .

  26A, 26B, and 26C show the light receiving surfaces of a three-plate solid-state electronic image sensor. From the three-plate solid-state electronic image sensor, the subject image of the red signal component shown in (A), the subject image of the green signal component shown in (B), and the blue signal component shown in (C) from each solid-state electronic image sensor. Each subject image is obtained. A face image area can be determined by performing face image detection processing as shown in FIG. 16 using two of the obtained three subject images.

  FIG. 27A shows a light receiving surface of a visible light sensor (imaging device), and FIG. 27B shows a light receiving surface of an infrared sensor.

  Referring to (A), in the visible light sensor, R filters, G filters, and B filters are regularly arranged at a constant period, as in the single-plate image sensor shown in FIG. With reference to (B), the infrared sensor is provided with an IR filter that transmits light components in the infrared region. (A) A visible light image represented by a video signal output from a visible light sensor having a light receiving surface and a red represented by a video signal output from a visible light sensor having a light receiving surface illustrated in (B). A face image area can also be determined by performing face image detection processing as shown in FIG. 16 using two subject images with an external light image.

  FIG. 28 shows a light receiving surface of the visible light / infrared sensor.

  On the light receiving surface of the visible light / infrared sensor, R filters, G filters, B filters, and IR filters are regularly arranged at a constant period. Subject image represented by red component video signal, subject image represented by green component video signal, subject image represented by blue component video signal and infrared component image from visible light / infrared sensor A subject image represented by the signal is obtained. A face image area can also be determined by performing face image detection processing as shown in FIG. 16 using two of the four subject images.

  As described above, the first and second subject images are obtained by imaging the subject using different imaging devices (imaging devices and sensors), and when the face image area is detected, the first subject image is detected. A first score of face image likelihood obtained in the face image detection process performed on the subject image and a second score of face image likelihood obtained in the face image detection process performed on the second subject image The face image area may be determined based on the sum of the first score, the face image area may be determined by weighting the sum of the first score and the second score, and the first score and the second score may be determined. The face image area may be determined based on the sum of the weighted scores by weighting the scores of 2. Of course, the face image area may be determined by combining these face image area determination processes.

  FIGS. 29 to 31 are time charts showing processing timings in the face image detection processing. By subjecting the solid-state electronic image sensor to interlace scanning, a subject image in the first field and a subject image in the second field are obtained alternately. In either case, the subject is imaged by performing exposure between times t1 and t2. Then, the video signal representing the subject image in the first field is read out between times t2 and t3.

  The face image area can be determined by performing face image detection processing as shown in FIG. 16 using the obtained subject image of the first field and subject image of the second field.

  In FIG. 29, face image detection processing is performed in order from the subject image in the field for which reading has been completed.

  Reading of the video signal of the second field is started from time t2, and face image detection processing of the subject image of the first field is performed. The reading of the video signal of the second field ends at time t4, and the face image detection process of the first subject image ends at time t5. The face image detection process of the second subject image ends from time t5 after the end of reading the video signal of the second field. The time until the end of the face image detection processing of the two subject images is shortened.

  In FIG. 30, face image detection processing is sequentially performed after the reading of the video signals of the subject images in all fields is completed.

  When the reading of the video signal of the first field is completed at time t3, the reading of the video signal of the second field is started and ends at time t4. The face image detection process of the first subject image is started from time t4 when both the reading of the video signal of the first field and the reading of the video signal of the second field are completed, and ends at time t5. The face image detection process of the second subject image is started from time t5, and ends at time t6. Since the reading process and the face image detection process are performed in different time zones, the burden on the processing apparatus is reduced.

  In FIG. 31, the face image detection of two subject images is performed simultaneously after the reading of the video signals of the subject images of all fields is completed.

  By the time t4, the reading of the video signal of the first field and the reading of the video signal of the second field are finished in order. Between time t5 and t6, the face image detection of the first subject image and the face image detection of the second subject image are performed in parallel.

  FIG. 32 is a flowchart showing a processing procedure of face image detection processing from the subject image in the first field and the subject image in the second field.

  The subject image of the first field and the subject image of the second field are acquired (steps 71 and 72), the face image detection processing of the subject image of the first field (step 73), and the face image detection processing of the subject image of the second field (Step 74) is performed.

  The detection result obtained by the face image detection process is collated (step 75), and the detection result reflecting the collation result is registered (step 76).

  FIG. 33 shows another embodiment and shows a subject image obtained by so-called bracket imaging.

  When the subject is imaged with the exposure amount set to a relatively small E1, a slightly dark first subject image α21 is obtained (darkness is represented by hatching). The first subject image α21 includes a person image S30. By performing the face image detection process on the first subject image α21, the face image portion of the person image S30 is detected as the face image region C33. A face image area C34 is also detected.

  Subsequently, the same subject is imaged with the exposure amount set to E2, which is relatively high, so that a bright first subject image α22 is obtained. The second subject image α22 also includes a person image S35 corresponding to the person image S30 of the first subject image α21. By performing image detection processing on the second subject image α22, the face image portion of the person image S35 is detected as the face image region S35. A face image area C36 is also detected.

  By comparing the detection result of the face image area of the first subject image α21 with the detection result of the face image area of the second subject image α22, the face image region C35 of the second subject image α22 is a face. It is determined as an image area. It will be understood that the face image region C34 detected in the first subject image α21 and the face image region C36 detected in the second subject image α22 are not treated as face image regions.

  FIG. 34 shows a subject image or the like obtained by so-called bracket imaging when the subject includes a dark part.

  When the subject is imaged with the exposure amount set to relatively small E1, the first subject image α23 that is slightly darker as described above is obtained. The first subject image α23 includes a person image S32 and a person image S33 present in the shade of a tree. By performing face image detection processing on the first subject image α23, the face image portion of the person image S32 is detected as the face image area S32. However, since the person image S33 is in the shade of the tree, the face image of the person image S33 is detected. The portion is not detected as a face image area.

  When the same subject is imaged with the exposure amount set to E2, which is relatively high, a bright second subject image α24 is obtained. The second subject image α24 includes person images S34 and S35 corresponding to the person images S32 and S33 included in the first subject image α23. Since the portion where the person image S35 exists is not so dark, the face image detection processing is performed on the second subject image α24, so that the face image portion of the person image S34 and the face image portion of the person image S35 are respectively face images. Regions C38 and C39 are detected.

  The above-described detection result collation process is stopped, and all the detected face image areas C37, C38 and C39 are added, so that the face image area not detected in the first subject image α23 becomes the face image area C39. Can be detected as

  FIG. 35 shows a subject image or the like obtained by changing signal processing.

  The first subject image α25 includes a large amount of noise by enlarging the contour enhancement process. The first subject image α25 includes a person image S36, and a face image portion is detected as a face image region C40 by performing face image detection processing. A face image area C41 is also detected.

  The second subject image α26 has less noise by reducing the contour enhancement process. The second subject image α26 also includes a person image S37 corresponding to the person image S36 included in the first subject image α25. By performing the face image detection process on the second subject image α26, the face image portion of the person image S37 is detected as the face image region C42. A face image area C43 is also detected.

  The face image region C42 is determined as the face image region by collating the detection result of the face image of the first subject image α25 with the detection result of the second subject image α26.

  FIG. 36 is a flowchart showing a processing procedure when face image detection processing is performed on two subject images obtained by different signal processing.

  The subject is imaged (step 81), and the first signal processing is performed on the subject image obtained by the imaging (step 82), and the second signal processing is performed on the subject image. As a result, a second subject image is obtained (step 83). A face image detection process for the first subject image is performed (step 84), and a face image detection process for the second subject image is performed (step 85).

  The collation process of the face image detection result of the first subject image and the face image detection result of the second subject image is performed (step 86), and the detection result obtained as a result of the collation is registered (step 87).

  FIG. 37 to FIG. 44 are flowcharts showing the processing procedure using the detected face image.

  FIG. 37 and FIG. 38 are flowcharts showing a processing procedure for performing automatic exposure adjustment using the detected face image area.

  Referring to FIG. 37, the detection result of the face image is read (step 91), and the presence / absence of the face image area is determined (step 92).

  If there is a face image area (YES in step 92), the brightness of the image in the detected face image area is calculated (step 93). If there are a plurality of face image areas (YES in step 94), the brightest face image area is selected (step 95). Face-weighted multi-pattern photometry is performed in which a photometric value is calculated based on the image in the selected face image area (step 96). If there is only one face image area, face-weighted multi-pattern metering is performed based on the image in the face image area (step 96).

  If there is no face image area (NO in step 92), center-weighted multi-pattern photometry is performed in which a photometric value is calculated based on the central portion (step 97).

  FIG. 38 shows another processing procedure of automatic exposure adjustment. In FIG. 38, the processing shown in FIG.

  In the process shown in FIG. 38, when there are a plurality of face image areas (YES in step 101), the one having the largest face image area is selected (step 102). Photometry is performed based on the image in the selected face image area (step 96).

  39 and 40 are flowcharts showing a processing procedure for performing automatic focusing control using the detected face image area.

  Referring to FIG. 39, the face image detection result is read (step 111), and if there is a face image area (YES in step 112), it is confirmed whether there are a plurality of face image areas (YES in step 113). If there are a plurality of face image areas (YES in step 113). The largest face image area is selected (step 114), and the image in the selected face image area is set as the focus area (step 115). If there is only one face image area (NO in step 113), that one face image area is set as the in-focus area (step 115). Automatic focusing processing is performed so that the set area is in focus (step 117).

  If there is no face image area (NO in step 112), automatic focusing processing is performed so that the center portion of the subject image is in focus (step 116).

  FIG. 40 shows another example of automatic focusing control. In FIG. 40, the same processes as those in FIG.

  If there are a plurality of face image areas (YES in step 113), the face image area of the subject image closest to the center is selected (step 118).

  41 and 42 are flowcharts showing the tone correction processing.

  Referring to FIG. 41, the detection result of the face image area is read (YES in step 121), and if there is a face image area (YES in step 122), the brightness of the image in each face image area is calculated ( Step 123). If there are a plurality of face image areas (YES in step 124), the brightest face image area is selected (step 125). If there is only one face image area (NO in step 124), that face image area is set. A gradation correction process is performed based on the image in the selected or set face image area (step 127). If there is no face image area (NO in step 122), predetermined gradation correction processing is performed (step 126).

  FIG. 42 shows another process of the gradation correction process. In FIG. 42, the same processes as those shown in FIG.

  If there are a plurality of face image areas (YES in step 124), the largest face image area is selected (step 128). Tone correction is performed using the image in the selected face image area (step 129).

  43 and 44 are flowcharts showing the color correction process.

  Referring to FIG. 43, the face image area is read (step 131), and it is confirmed whether there is a face image area (step 132). If there is a face image area (YES in step 132) and there are a plurality of face image areas (YES in step 133), the largest face image area is selected (step 144). If there is only one face image area (NO in step 133), that face image area is set. Color information is acquired from the image in the selected or set face image area (step 145), and color correction is performed so that an image of a desired color is obtained (step 147). If the face image area does not exist (NO in step 132), predetermined color correction is performed (step 146).

  FIG. 44 shows another example of color correction processing. In FIG. 44, the same processes as those shown in FIG.

  If there is a face image area (YES in step 132), the color information of the face image area is acquired (step 148). If there are a plurality of face image areas, the face image area having the color closest to the desired color is selected (step 149). If there is only one face image area (NO in step 133), that face image area is set. Color correction is performed so that the selected face image area or the image in the selected face image area has a desired color (step 147).

  In the above embodiment, the face image is detected. However, an image other than the face image, for example, an eye image may be determined in the same manner.

  In addition to automatic exposure control based on the face image area determined as described above, flash (strobe) control, power saving control, zoom amount control, and gain adjustment are performed based on the determined face image area. , White balance adjustment, contour enhancement, noise reduction, and the like may be performed.

2 Control circuit 3 CPU
6 Strobe control circuit 7 Strobe light emitting device
10 Image sensor
17 Face detection processing circuit

Claims (4)

First detecting means for detecting a first position where the target image exists from a first subject image among a plurality of subject images obtained by continuously capturing subjects that can be regarded as substantially the same;
Second detection means for detecting a second position where the target image is present from a second subject image among the plurality of subject images;
Based on the first position detected by the first detection means and the second position detected by the second detection means, at least one of the first subject image and the second subject image. Determining means for determining a position where the target image is present in the subject image, and the size of the first subject image is different from the size of the first subject image, and the size of the first subject image. And the second position detected by the second detection means and the second position detected by the second detection means so as to be the position when the size of the second subject image is the same. A position converting means for converting at least one of the positions;
The determining means determines a position where the target image exists based on the first position and the second position converted by the position converting means,
An imaging device for imaging a subject and outputting image data representing the subject image;
In response to the pre-imaging command, a first imaging control means for controlling the imaging device to image the subject and output first image data representing the first subject image, and in response to the main imaging command , Further comprising second imaging control means for controlling the imaging device so as to image the subject and output second image data representing the second subject image;
The first detection means includes the target image from among first subject images represented by first image data output from the imaging apparatus under the control of the first imaging control means. To detect the position to be
The second detection means includes the target image from the second subject image represented by the second image data output from the imaging device under the control of the second imaging control means. Is to detect the position to
Target image position detection device. A first detection unit that detects a first position where the target image exists from a first subject image among a plurality of subject images obtained by continuously capturing subjects that can be regarded as substantially the same;
A second detecting means for detecting a second position where the target image exists from a second subject image among the plurality of subject images;
Based on the first position detected by the first detection means and the second position detected by the second detection means, the determination means determines the first subject image and the second subject. Determining a position where the target image exists in at least one subject image of the image;
The position conversion means is such that the size of the first subject image is different from the size of the second subject image, and the size of the first subject image is the same as the size of the second subject image. Converting at least one of the first position detected by the first detection means and the second position detected by the second detection means to be a position in a certain case;
The determining means determines a position where the target image exists based on the first position and the second position converted by the position converting means,
The imaging device images the subject and outputs image data representing the subject image,
A first imaging control unit that controls the imaging device to image a subject and output first image data representing the first subject image in response to a pre-imaging command;
A second imaging control unit that controls the imaging device to image a subject and output second image data representing the second subject image in response to the main imaging command;
The first detection means includes the target image from among first subject images represented by first image data output from the imaging apparatus under the control of the first imaging control means. To detect the position to be
The second detection means includes the target image from the second subject image represented by the second image data output from the imaging device under the control of the second imaging control means. Is to detect the position to
A method for detecting the position of a target image. A computer-readable program for controlling a computer of a target image position detection device,
A first position where the target image exists is detected from a first subject image among a plurality of subject images obtained by continuously capturing subjects that can be regarded as substantially the same;
A second position where the target image exists is detected from a second subject image among the plurality of subject images;
Based on the detected first position and the detected second position, a position where the target image exists in at least one of the first subject image and the second subject image is determined. Let
A position where the size of the first subject image is different from the size of the second subject image, and the size of the first subject image is the same as the size of the second subject image; And converting at least one of the detected first position and the detected second position,
Based on the converted first position and second position, the position where the target image exists is determined;
In response to the pre-imaging command, the imaging device is controlled to image a subject and output first image data representing the first subject image,
In response to the main imaging command, the imaging device is controlled to image a subject and output second image data representing the second subject image,
In the first position detection process, the position where the target image exists is detected from the first subject image represented by the first image data output from the imaging device,
In the second position detection process, the position where the target image exists is detected from the second subject image represented by the second image data output from the imaging device.
A program for controlling a computer of a target image position detection apparatus.   A recording medium storing the program according to claim 3.

Images