JP5136705B2 - Printing device - Google Patents

Description

The present invention relates to a printing apparatus, a printing apparatus that detect an object, particularly included in the image.

  A face image included in the image data is detected, and using the detection result, printing processing, correction processing, and operation control of the image input device are performed. By moving the detection window to any position in the image data and determining whether a face image exists in the detection window, a face image existing at an unspecified position in the image data can be detected. (See Patent Document 1) In this document (see FIG. 6), the detection window is moved in order from the corner of the image data.

JP 2007-48108 A

  Since the processing load for determining whether or not a face image exists in the detection window is not small, moving the detection window to any position in the image data increases the processing load for the entire detection process. was there. If detection windows are set for all positions (pixels) of image data, omission of detection can be prevented, but the processing load increases remarkably. On the contrary, if the detection window is set only for a part of the position (pixels) of the image data, the processing load can be reduced, but a detection failure occurs.

The present invention has been made in view of the above problems, efficiently, and aims to provide a printing apparatus that detect the object without leaking.

  In order to solve the above problem, the present invention performs the following procedure when detecting an object from an image indicated by image data having position information about at least the first coordinate axis and the second coordinate axis. First, the detection window is sequentially moved sequentially by a first movement interval with respect to the first coordinate axis direction, and the detection window is sequentially moved by a second movement interval different from the first movement interval with respect to the second coordinate axis direction. . Further, each time the detection window moves to each position, the presence or absence of the object is determined for the image in the detection window. That is, while moving the detection window with respect to each of the first coordinate axis direction and the second coordinate axis direction, it is determined whether or not an object exists for the detection window.

  Here, when the detection window is moved in each of the first coordinate axis direction and the second coordinate axis direction, the movement interval in the first coordinate axis direction and the movement interval in the second coordinate axis direction are set to different sizes. By doing so, it is possible to detect an object with no omission in one coordinate axis direction, and it is possible to detect an object with less processing load in the other coordinate axis direction. That is, by setting the optimal movement interval according to each direction, it is possible to perform object detection without leakage and efficient. The object only needs to have a specific image characteristic. For example, a face of a person / animal or a facial organ can be detected.

  Furthermore, the movement positions of the detection windows adjacent to each other in the second coordinate axis direction may be shifted by a predetermined amount in the first coordinate axis direction. Accordingly, it is possible to prevent the movement positions of the plurality of detection windows adjacent to each other in the second coordinate axis direction from being arranged at the same position in the first coordinate axis direction, and to perform object detection without leakage. The amount of deviation in the first coordinate axis direction may be, for example, half the first movement interval.

  Further, various coordinate axis systems can be applied as the first coordinate axis and the second coordinate axis. For example, the first coordinate axis direction and the second coordinate axis direction may be the vertical direction and the horizontal direction of the image indicated by the image data, respectively. In this way, different movement intervals can be set in the vertical direction and the horizontal direction. Furthermore, the first coordinate axis direction and the second coordinate axis direction may be set to a radial direction and a circumferential direction of a circle centered on a predetermined position of the image indicated by the image data.

  Furthermore, the technical idea of the present invention can be implemented not only by a specific object detection method but also by an object detection apparatus. That is, the present invention can also be specified as an object detection apparatus having means corresponding to each step performed by the object detection method described above. Of course, when the above-described object detection apparatus reads a program to realize each of the above-described means, the technical features of the present invention can be applied to a program for executing a function corresponding to each of the means and various recording media on which the program is recorded. It goes without saying that the idea can be embodied. Needless to say, the object detection device of the present invention can be distributed not only by a single device but also by a plurality of devices. Further, the object detection method of the present invention may be realized in a printing apparatus such as a printer or an image input apparatus such as a digital still camera.

It is a block diagram which shows the hardware constitutions of an image processing apparatus. It is a block diagram which shows the software structure of an image processing apparatus. It is a flowchart which shows the flow of an image process. It is a flowchart which shows the flow of a face detection process. It is a figure which shows an example of work image data. It is a graph which shows the relationship between a size counter and size. It is a figure which shows the relationship between a detection window and work image data. It is a graph which shows the relationship between a rotation angle counter and the rotation angle of a detection window. It is a figure which shows a mode that the position of a detection window moves. It is a flowchart which shows the flow of a face determination process. It is a figure which shows a mode that a feature-value is calculated from window image data. It is a figure which shows an example of the structure of a neural network. It is a figure which shows typically a mode that a neural network is learned. It is a flowchart which shows the flow of a skin color adjustment process. It is a flowchart which shows the flow of the face detection process concerning a modification. It is a figure which shows the relationship between the radial direction counter concerning the modification, and the circumferential direction counter. It is a figure which shows a mode that the position of a detection window moves in a modification.

Hereinafter, embodiments of the present invention will be described in the following order.
1. Configuration of image processing device:
2. Image processing:
2-1. Face detection process:
2-2. Face judgment processing:
2-3. Skin tone adjustment and printing process:
3. Variations:
3-1. Modification 1:

1. Configuration of Image Processing Device FIG. 1 shows the configuration of a computer that specifically implements an image processing device according to an embodiment of the present invention. In FIG. 1, a computer 10 includes a CPU 11, a RAM 12, a ROM 13, a hard disk drive (HDD) 14, a general purpose interface (GIF) 15, a video interface (VIF) 16, an input interface (IIF) 17, and a bus 18. The bus 18 implements data communication between the elements 11 to 17 constituting the computer 10, and communication is controlled by a chip set (not shown). The HDD 14 stores program data 14a for executing various programs including an operating system (OS), and the CPU 11 executes calculations according to the program data 14a while expanding the program data 14a in the RAM 12. .

  Further, the HDD 14 stores image data 14b input by a digital still camera or a scanner. The GIF 15 provides an interface conforming to the USB standard, for example, and connects an external printer 20 to the computer 10. The VIF 16 connects the computer 10 to an external display 40 and provides an interface for displaying an image on the display 40. The IIF 17 connects the computer 10 to an external keyboard 50a and a mouse 50b, and provides an interface for the computer 10 to acquire input signals from the keyboard 50a and the mouse 50b.

  FIG. 2 shows a software configuration of a program executed in the computer 10. In the figure, an operating system (OS) PG1, an image processing application PG2, and a printer driver PG3 are executed. The OS PG1 provides an interface between the programs, and the printer driver PG3 executes a process for controlling the printer 20. The image processing application PG2 includes a work image acquisition unit PG2a, a detection window setting unit PG2b, a window image acquisition unit PG2c, a feature amount calculation unit PG2d, a face determination unit PG2e, and an image quality adjustment unit PG2f. Details of processing executed by each of the modules PG2a to PG2f constituting the image processing application PG2 will be described together with a flow of image processing described later.

2. Image Processing FIG. 3 shows the overall flow of image processing. In step S100, the work image acquisition unit PG2a displays a predetermined UI screen on the display 40 and accepts an operation from the keyboard 50a and the mouse 50b. Thereby, the image data 14b that the user wants to print is acquired from among the plurality of image data 14b stored in the HDD 14 or the like. The work image acquisition unit PG2a converts the resolution of the image data 14b to generate work image data WD of 320 pixels × 240 pixels (QVGA). In step S200, the detection window setting unit PG2b executes face detection processing, and sequentially sets detection windows SW in the work image data WD. Each time the detection window SW is set, in step S400, the feature amount calculation unit PG2d and the face determination unit PG2e cause the face determination process for the detection window SW to be executed, and the determination result is output. In accordance with the determination result, the image quality adjustment unit PG2e executes the flesh color adjustment process in step S500, and outputs the image data 14b after the image quality adjustment to the printer driver PG3. Then, the printer driver PG3 executes print control processing, and the printer 20 prints the image data 14b after the image quality adjustment.

2-1. Face Detection Processing FIG. 4 shows the flow of face detection processing, and FIG. 5 shows work image data WD. In FIG. 5, a detection window (detection area) SW of a square area is set in the work image data WD, and the detection window setting unit PG2b moves the detection window SW in the work image data WD without fail by face detection processing. Note that the longitudinal direction of the work image data WD is represented as the x-axis (first coordinate axis), and the short direction is represented as the y-axis (second coordinate axis). The origin of the x-axis and y-axis is the upper left corner of the work image data WD. Position. When moving the detection window SW, the size of the detection window SW and the rotation angle of the detection window SW in the image plane are sequentially shifted. In order to shift the detection window SW as described above, it is necessary to sequentially shift the following parameters.
-Detection window SW size parameter: S
-Parameter for the center position of the detection window SW: P
-Parameter of rotation angle of detection window SW: T
In step S210, the detection window setting section PG2b each counter _{n s, n x, n y} , and resets the n _t. Each counter _ns , _nx , _ny , _nt is an integer value for sequentially shifting the above-described parameters S, P, T, and is set to 0 by reset. In step S220, 1 is added to the size counter n _s.

Figure 6 shows the relationship between the size S of the size counter n _s and the detection window SW. In the figure, it is shown that the size S (vertical and horizontal lengths) of the detection window SW gradually decreases as the size counter _ns increases. In this embodiment, the size counter n _s and the size S have a linear relationship, and each time the size counter n _s increases by 1 between 1 to 15, the size S of the detection window SW (vertical and horizontal) The length is reduced by 12 pixels. The maximum size S when the size counter _ns is 1 is 200 pixels, which is slightly shorter than the short length of the detection window SW, and the minimum size S of the detection window SW when the size counter _ns is 15 is obtained. 20 pixels. The relationship between the size counter n _s and the size S of the detection window SW shown here is an example, and these may have a non-linear relationship, or the inclination, intercept, etc. may be changed. In the next step S230, S240, in order to set the parameters P of the center position of the detection window SW, adds 1 to the y direction counter n _y and x direction counter n _x.

上記の（１）式において、ｄ_x，ｄ_yは検出窓ＳＷの中心位置Ｐの各方向への単位移動距離（画素数）を示す一定の移動間隔を表しており、移動間隔ｄ_x，ｄ_yと方向カウンタｎ_x，ｎ_yをそれぞれ乗算することにより、検出窓ＳＷの中心位置Ｐのｘ，ｙ座標を算出する。移動間隔ｄ_x，ｄ_yは検出窓ＳＷの縦および横の長さを示すサイズＳに基づいて下記の（２）式によって算出される。

上記の（２）式が示すように、移動間隔ｄ_x，ｄ_yは互いに異なる値となっており、ｐ_x＞ｐ_yとなっている。上記の（２）式によって移動間隔ｄ_y＜１となる場合は、ｄ_y＝１とする。なお、ｘ方向カウンタｎ_xが取り得る範囲は１〜３２０／ｄ_x（整数部分）とし、ｙ方向カウンタｎ_yは１〜２４０／ｄ_y（整数部分）とする。ｙ方向カウンタｎ_yが偶数である場合には、検出窓ＳＷの中心位置Ｐのｘ，ｙ座標を下記の（３）式によって算出する（ステップＳ２５０ｂ）。 In step S245, y direction counter n _y determines whether is odd or even. Here, when the y-direction counter n _y is an odd number, the x and y coordinates of the center position P of the detection window SW are calculated by the following equation (1) (step S250a).

In the above (1), d _x, d _y represents a certain movement distance showing a unit moving distance (number of pixels) in each direction of the center position P of the detection window SW, the movement distance d _x, d by multiplying _y and direction counter n _x, n _y, respectively, x of the center position P of the detection window SW, calculates the y-coordinate. Movement distance d _x, d _y is calculated by the following equation (2) based on the size S that indicates the length of sides of the detection window SW.

As indicated by the equation (2), the movement distance d _x, d _y is a different value, and has a p _x> p _y. When the movement interval d _y <1 by the above equation (2), d _y = 1. The range that can take the x direction counter n _x is a 1 to 320 / d _x (integer part), the y direction counter n _y and to 240th / d _y (integer part). If y direction counter n _y is even, x of the center position P of the detection window SW, the y coordinate is calculated by the following equation (3) (step S250b).

上記の（３）式によれば、ｙ方向カウンタｎ_yが偶数である場合は、ｙ方向カウンタｎ_yが奇数である場合よりも、検出窓ＳＷの中心位置Ｐのｘ座標がｘ方向の移動間隔ｐ_xの半分だけ進んだ位置となる。ステップＳ２５０ａ，２５０ｂにおいては、それぞれ以上のようにして算出した中心位置Ｐを中心として、サイズＳの検出窓ＳＷを生成する。ステップＳ２６０においては、検出窓ＳＷの一部が作業画像データＷＤの外側にはみ出ていないか否かを判定する。

According to the above equation (3), when the y-direction counter _ny is an even number, the x coordinate of the center position P of the detection window SW moves in the x direction more than when the y-direction counter _ny is an odd number. the advanced position just half of the interval p _x. In steps S250a and 250b, a detection window SW of size S is generated around the center position P calculated as described above. In step S260, it is determined whether a part of the detection window SW does not protrude outside the work image data WD.

FIG. 7 schematically shows the state of determination in step S260. As shown in the drawing, each counter n _s, n _x, the position and size of the detection window SW by variations in n _y shifts, there counter n _s, n _x, detected at the combination of n _y window SW A part of the image data protrudes outside the work image data WD. As it is impossible to perform a normal face detection in such a case, the counter n _s, n _x, to skip the face detection processing for the detection window SW specified by n _y. On the other hand, if all of the detection window SW enters the inside of the working image data WD is, it adds 1 to the rotation angle counter n _t for the detection window setting section PG2b in step S270 to set the rotation angle T of the detection window SW To do.

FIG. 8 shows the relationship between the rotation angle counter n _t and the rotation angle T of the detection window SW. As shown in the figure, when the rotation angle counter _nt is 1, the upward direction of the work image data WD is on the detection window SW, and when the rotation angle counter _nt is 2, the right direction of the work image data WD is the detection window SW. When the rotation angle counter _nt is 3, the downward direction of the work image data WD is above the detection window SW. When the rotation angle counter _nt is 4, the left direction of the work image data WD is above the detection window SW. And That is, in this embodiment, the rotation angle counter n _t is the be rotated by 90 degrees detection window SW clockwise every increase 1. Note that the rotation angle of the detection window SW is not limited to 90 degrees, and may be, for example, 30 degrees.

As described above, with the exception of the case where skipped at step S260, all the counters _{n s, n x, n y} , n t can be the size and location and rotation angle detection window setting of the detection window SW setting This is uniquely specified by the part PG2b. In step S280, the window image acquisition unit PG2c extracts the image data inside the currently determined detection window SW and acquires it as window image data XD. The acquired window image data XD is output to the feature amount calculation unit PG2d, and the face determination process for the window image data XD is executed in step S400. Details of the face determination process will be described later.

When the window image data XD is output to the feature amount calculation unit PG2d, the detection window setting unit PG2b determines whether or not the rotation angle counter n _t = 4 in step S290. That is, it is determined whether or not the setting of the rotation angles in the four directions is completed at the current size S and center position P of the detection window SW. If the rotation angle counter n _t <4, the process returns to step S270, and 1 is added to the rotation angle counter n _t . As a result, the detection window SW rotated 90 degrees clockwise can be set next time. On the other hand, if the rotation angle counter n _t = 4, the rotation angle counter n _t is reset to 0 in step S295, and whether or not the _x -direction counter n _x = 320 / d _x (integer part) is determined in step S300. Determine. That is, it is determined whether or not the setting of the detection window SW is completed for all the positions in the x direction in the current size S of the detection window SW and the position in the y direction. when x direction counter n _x <320 / d _x (integer part), the process returns to step S240, 1 is added to the x direction counter n _x. As a result, next time, the detection window SW can be set at the center position P advanced to the right by the movement interval p _x (6 pixels) in the x direction.

On the other hand, if x direction counter n _x = 320 / d _x (integer part), x direction counter n _x is reset to 0 in step S305, and y direction counter n _y = 240 / d _y (integer) in step S310. Part) is determined. That is, it is determined whether or not the setting of the detection window SW has been completed for all the center positions P in the current size S of the detection window SW. If the y-direction counter n _y <240 / d _y (integer part), the process returns to step S230, and 1 is added to the y-direction counter n _y . Thereby, the detection window SW can be set at the center position P advanced by one pixel (movement interval p _y ) in the y direction next time. On the other hand, if the y-direction counter n _y = 240 / d _y (integer part), the y-direction counter n _y is reset to 0 in step S315, and whether or not the size counter n _s = 15 is determined in step S320. judge. That is, it is determined whether or not the detection window SW has been set for all the sizes S. If the size counter n _s <15, the process returns to step S220, and 1 is added to the size counter n _s . As a result, the detection window SW having a size S that is slightly smaller the next time can be set. On the other hand, if the size counter n _s = 15, it is determined that the setting of the detection window SW has been completed for all combinations of parameters (size S, position P, and rotation angle T), and the process ends.

  In the face detection process described above, the window image data XD is characterized every time one of the size S, position P, and rotation angle T of the detection window SW is updated, except when skipped in step S260. The face determination process for the window image data XD is output to the amount calculation unit PG2d and executed in step S400. In this face determination process, since it is determined whether or not a face image exists in the window image data XD, the presence or absence of a face image is determined based on the size S, position P, and rotation angle T at which the detection window SW is set. I can go.

FIG. 9 shows how the position of the detection window SW moves in the present embodiment. As shown in the figure, it is possible to work starting from the upper left corner of the image data WD, gradually moves the center position P of the detection window SW to the right with increasing x direction counter n _x. When the x-direction counter n _x = 320 / d _x (integer part), the center position P of the detection window SW reaches the right end of the work image data WD. When reaching the right side, the x-direction counter _nx is reset to 0 and 1 is added to the y-direction counter _ny , so that the center position P of the detection window SW moves to a position one step below the left end. . When the above movement is repeatedly executed, the center position P of the detection window SW is shifted by d _x / 2 in the x direction depending on whether the y direction counter _ny is an odd number or an even number. By doing so, it is possible to alleviate the difference in density of the position to which the center position P of the detection window SW is moved without increasing the number of face determination processes to be described later. No face detection is possible. Further, according to the above equation (2), the density in the y direction of the center position P of the detection window SW is five times the density in the x direction. In this way, by making the density of the center position P of the detection window SW nonuniform in the x direction and the y direction, face determination processing in which the determination characteristics (robustness) are nonuniform in the x direction and the y direction (step S400). ) Can be realized.

2-2. Face Determination Processing FIG. 10 shows the flow of face determination processing. When the window image data XD is acquired in step S410, the feature amount calculator PG2d calculates a plurality of feature amounts from the window image data XD in the next step S420. As for the feature amount, various types of filters are applied to the window image data XD, and the feature amount (average value, maximum value, minimum value, standard deviation, etc.) indicating the state of brightness average, edge amount, contrast and the like in the filter is included. It is obtained by calculating. Note that the window image data XD has different sizes depending on the size S of the detection window SW, but the resolution is converted into a predetermined size in advance in step S410.

  FIG. 11 shows how the feature amount is calculated from the window image data XD. In the figure, a large number of filters FT are prepared for the window image data XD, and each filter FT is sequentially applied to the window image data XD. For example, twelve feature values CA1 to CA12 for images in each filter FT. Are calculated sequentially. When the feature amounts CA1 to CA12 can be calculated, in step S430, the face determination unit PG2e inputs the feature amounts CA1 to CA12 to the neural network NN prepared in advance, and a determination result indicating whether or not a face image exists as the output is obtained. calculate. Note that the feature amount CA is not limited as long as the image feature indicated by the window image data XD is indexed, and the filter FT is not necessarily used.

  FIG. 12 shows an example of the structure of the neural network NN. The neural network NN has a basic structure in which the value of the unit U in the subsequent layer is determined by a linear combination of the values of the unit U in the previous layer (the suffix i is the identification number of the unit U in the previous layer). Further, the value obtained by the linear combination may be used as the value of the unit U of the next layer as it is, but the value obtained by the linear combination is converted by a non-linear function such as a hyperbolic tangent function, for example. By determining the value of U, non-linear characteristics may be provided. The neural network NN of this embodiment has a three-layer structure, and the neural network NN is composed of one intermediate layer, an input layer, and an output layer. Each feature amount CA1 to CA12 can be input to the input layer of the neural network NN, and the output value K (value normalized to 0 to 1) can be output from the output layer. If the output value K is 0.5 or more, it is determined that a face image exists in the window image data XD, and if the output value K is less than 0.5, it is determined that no face image exists in the window image data XD. (Step S440). If such a neural network NN is prepared in the HDD 14 in advance, whether or not a face image exists in the window image data XD can be determined based on the feature amounts CA1 to CA12.

  FIG. 13 schematically shows how the neural network NN is learned. In the present embodiment, learning is performed by the error back propagation method, so that the number of units U, the size of weight w and the value of bias b at the time of linear combination between units U are optimized. Is done. In learning by the back propagation method, first, the magnitude of the weight w and the value of the bias b at the time of linear combination between the units U are initially set to appropriate values.

  The feature amounts CA1 to CA12 are calculated for learning image data for which it is known whether or not a face image exists, in the same procedure as in step S420, and the feature amounts CA1 to CA12 are set in the initially set neural network NN. An input value K is obtained. It is necessary to prepare as much learning image data as possible, and it is necessary to prepare various races, genders, ages, and the like so that face images in various states can be detected. Furthermore, the face image included in the image data 14b photographed by a digital still camera or the like may be oriented in various directions. Therefore, learning image data including face images oriented in various directions is prepared. In addition, since there is a high possibility that the face is photographed with the face facing left and right rather than the face facing up and down, a larger number of learning image data photographed with the face facing left and right are prepared.

  In this embodiment, it is desirable that 1 is output as the output value K for learning image data in which a face image exists, and 0 is output as the output value K for learning image data in which no face image exists. Is desirable. However, since the weight w and the value of the bias b at the time of linear combination between the units U are merely set to appropriate values, there is an error between the actual output value K and the ideal value. Will occur. The weight w and the bias b for each unit U that minimizes such an error are calculated using a numerical optimization method such as a gradient method. This error is propagated from the subsequent layer to the previous layer, and is optimized in order from the weight w and the bias b for the subsequent unit U. According to the neural network NN learned as described above, an output value K close to 1 can be obtained for the feature amounts CA1 to CA12 when a face image is present in the window image data XD. An output value K close to 0 can be obtained for the feature amounts CA1 to CA12 when no face image exists in the data XD. Therefore, it is possible to determine whether or not a face image exists in the window image data XD by performing threshold determination with an appropriate threshold 0.5 (step S440).

  If it is determined in step S440 that a face image exists for the window image data XD, the size S, position P, and rotation angle T of the detection window SW from which the window image data XD was obtained are stored in the RAM 12 (step S450). ), Return. On the other hand, if it is determined in step S430 that no face image exists for the window image data XD, the process directly returns. Since the window image data XD is sequentially output to the feature amount calculation unit PG2d by the face detection process described above, the processes of steps S410 to S440 are executed each time. As a result, as shown in FIG. 9, the presence or absence of a face image can be determined for each of the detection windows SW that are sequentially moved.

2-3. Skin Color Adjustment and Printing Process FIG. 14 shows the flow of the skin color adjustment process executed by the image quality adjustment unit PG2e. In step S320 of the face detection process described above, when it is determined whether or not the detection window SW has been set for all the sizes S, the face detection process ends and is executed in synchronization with the face detection process. The face determination process is also terminated. In step S510, the end of face detection processing and face determination processing is detected, and image data 14b to be adjusted is acquired. The image data 14 acquired here is the image data 14b that is the target of the face detection process and the face determination process. In step S520, the size S, position P, and rotation angle T of the detection window SW determined to have a face image are read from the RAM 12. In step S530, an area corresponding to the detection window SW where it is determined that a face image exists is specified in the image data 14b. Since the size S and position P of the detection window SW are acquired from the RAM 12, the area corresponding to the detection window SW can be specified by converting this to the image size of the image data 14b.

  In step S540, the color of the flesh color pixel included in the region specified in step S530 is sampled, and parameters for color tone adjustment for the image data 14b are set based on the color. Here, first, the skin color pixels included in the area specified in step S540 are specified based on the color value (for example, RGB value or HSV value) of each pixel, and the color value is corrected to a preferable skin color. An adjustment parameter is calculated. The adjustment parameter corresponds to a parameter for adjusting color balance, white balance, and brightness, for example. A color value preferable as a skin color is stored in the HDD 14 in advance, and an adjustment parameter is calculated so that the color value of each skin color pixel approaches the preferable color value. Since the area where the face image exists is specified in advance by the detection window SW, the adjustment parameter can be obtained based on the skin color pixels of the face image. When a plurality of detection windows SW that are determined to have face images are detected, an average adjustment parameter is calculated. When the adjustment parameter can be calculated, in step S545, the color value of each pixel is adjusted based on the adjustment parameter. When the skin color adjustment is completed as described above, the adjusted image data 14b is output to the printer driver PG3 in step S550. Then, the printer driver PG3 sequentially executes resolution conversion processing, color conversion processing, halftone processing, and rasterization processing on the image data 14b, and causes the printer 20 to print an image corresponding to the image data 14b after the image quality adjustment. Let

3-1. Modification 1
FIG. 15 shows the flow of face detection processing according to this modification. In step S230, S240 of this modification, instead, the center position of the detection window SW in the radial direction counter n _r and the circumferential counter n _a prior embodiment x direction counter was used in the form n _x and y direction counter n _y Is used to set the parameter P.

一方、径方向カウンタｎ_rが偶数である場合（ステップＳ２４５）には、検出窓ＳＷの中心位置Ｐのｘ，ｙ座標を下記の（５）式によって算出する。

このようにすることにより、径方向カウンタｎ_rが偶数である場合には、径方向カウンタｎ_rが奇数である場合よりも周方向に１５度進んだ位置に検出窓ＳＷの中心位置Ｐを設定することができる。 Figure 16 shows the parameters P of the center position of the detection window SW, the relationship between the radial counter n _r and the circumferential counter n _a. The radial direction counter n _r is associated with the center position P of the detection window SW in the radial direction with the center coordinate of the work image data WD as the origin. The radial counter n _r takes an integer value from 1 to 6. When the position of the central position P in the radial direction is represented by r, the position r is 0 when the radial counter n _r is 1, and is a quadratic curve (n _r = 1) having a maximum value when the radial counter n _r is 6. Is a quadratic curve that is convex downward. Note that the maximum value of the position r is slightly smaller than the distance from the origin to the four corners of the detection window SW. On the other hand, the circumferential direction counter n _a take integer values up to 12. When an angle formed by a straight line (right side) passing through the origin in the horizontal direction and a straight line connecting the origin and the center position P of the detection window SW is represented by θ, the angle θ is 0 degrees when the circumferential counter n _a is 1. Each time the circumferential counter n _a increases by 1, it rotates 30 degrees counterclockwise (θ = 30 * n _a ). Incidentally, the circumferential direction counter n _a for taking the integer value of up to 12, can be shifted only angle θ one turn (360 degrees). Note that the upper limit value of the circumferential counter n _a is 1 only when the radial counter n _r = 1. In step S250, based on the above definition, with the detection window setting section PG2b to specify positions r and the angle θ with respect to the radial direction of the center of the detection window SW from the radial direction counter n _r and the circumferential counter n _a, radially When the counter n _r is an odd number (step S245), the x and y coordinates of the center position P of the detection window SW are calculated by the following equation (4).

On the other hand, when the radial direction counter n _r is an even number (step S245), the x and y coordinates of the center position P of the detection window SW are calculated by the following equation (5).

In this way, when the radial counter n _r is an even number, the center position P of the detection window SW is set at a position advanced 15 degrees in the circumferential direction as compared with the case where the radial counter n _r is an odd number. can do.

FIG. 17 shows how the position of the detection window SW moves in this modification. As shown in the figure, starting from the center of the work image data WD, the center position P of the detection window SW can be moved outward as the radial counter _nr increases. In many image data 14b (working image data WD) photographed by a digital still camera or the like, the main subject person is composed at the center of the entire image, so that a face image can be detected at the initial stage of processing. For example, when the face detection process is terminated when a predetermined number of face images are detected in the face determination process described later, the process can be terminated earlier and the processing load can be reduced. . Further, since the radial position r corresponding to the radial counter n _r is represented by a quadratic curve convex with n _r = 1 at the top, the detection window SW is near the center of the work image data WD. The center position P can be made higher in density, and the center position P can be made lower in density as the distance from the center of the work image data WD increases. In this way, the detection window SW can be finely moved at a position where the face image is likely to be included, and the number of times the face determination process is performed at a position where the face image is unlikely to be included is reduced. Can be made. Furthermore, each time the center position P of the detection window SW moves one step outward, the position of the center position P of the detection window SW in the circumferential direction can be shifted by 15 degrees, thereby realizing face detection without leakage. Can do.

In the above description, the image processing method of the present invention is exemplified as being executed on a computer. However, the image processing method may be executed by an image device such as a printer, a digital still camera, or a scanner. If the image processing method of the present invention is performed by a printer, it is possible to execute image processing that approximates an ideal face image during printing. Of course, the printer is not limited to a home printer, and can be applied to a business photo print system. Further, the present invention can be realized not only in a printer that outputs image processing results on a printing sheet but also in an apparatus that outputs image processing results on a display such as a photo viewer. Further, if the image processing method of the present invention is performed with a digital still camera, it is possible to obtain a photographing result that is an ideal skin-colored face image. Furthermore, ATM (Automated Teller) that performs person authentication
The present invention can also be applied to a machine). The method is not limited to performing face determination using the neural network NN, and various determination methods in the feature amount space of the feature amount described above can also be used. For example, a support vector machine may be used, or a plurality of stages of determination devices with different determination boundary surfaces may be cascade-connected.

  DESCRIPTION OF SYMBOLS 10 ... Computer, 11 ... CPU, 12 ... RAM, 13 ... ROM, 14 ... HDD, 14a ... Program data, 14b ... Image data, 15 ... GIF, 16 ... VIF, 17 ... IIF, 18 ... Bus, 20 ... Printer, 40 ... display, 50a ... keyboard, 50b ... mouse, PG1 ... OS, PG2 ... image processing application, PG2a ... work image acquisition unit, PG2b ... detection window setting unit, PG2c ... window image acquisition unit, PG2d ... feature quantity calculation unit, PG2e: face determination unit, PG2f: image quality adjustment unit, PG3: printer driver.

Claims (3)

A printing apparatus that prints an image indicated by image data having position information about at least a first coordinate axis and a second coordinate axis,
  The detection window is sequentially moved periodically by a first movement interval with respect to the first coordinate axis direction, the detection window is sequentially moved by a second movement interval different from the first movement interval with respect to the second coordinate axis direction, and Detecting window setting means for shifting the movement positions of the detection windows adjacent to each other in the second coordinate axis direction by a predetermined amount in the first coordinate axis direction;
  Determination means for determining the presence or absence of an object for an image in the detection window each time the detection window moves to each position;
  And a printing unit that prints the image data that has been subjected to image processing according to the determination. The printing apparatus according to claim 1, wherein the predetermined amount is half the first movement interval. The first coordinate axis direction and the second coordinate axis direction are a radial direction and a circumferential direction of a circle centering on a predetermined position of an image indicated by the image data, respectively. The printing apparatus according to one item.

Images