JP5290734B2 - Noise reduction device and noise reduction method - Google Patents

Description

  The present invention relates to a noise reduction device and a noise reduction method.

  In an imaging apparatus such as a digital still camera or a video camera, image data generated based on an electrical signal output from an imaging device that converts optical information into an electrical signal includes noise, and thus image quality is degraded. Therefore, as a method for reducing noise, there is a case where filter processing is performed on image data.

  For example, Patent Document 1 discloses a technique for reducing noise in image data. In the technique of Patent Document 1, the noise amount is estimated in consideration of not only the signal level but also dynamically changing factors such as the image sensor temperature at the time of shooting and the analog gain according to the exposure time. The smoothing process is controlled according to the information on the pixels and neighboring pixels.

JP 2006-87030 A

  By the way, a subject photographed using an imaging device may include a plant such as a tree in a distant view. At this time, the leaves and flowers of the tree are very fine compared to the entire subject image. In an image in which a plant is photographed in a distant view, it is important that the plant part has a resolution that it looks like a real thing because the details are reliably expressed. However, distant view plant parts in the image have relatively high frequency components compared to other image parts. For this reason, when image processing is performed in the imaging apparatus or the like, it is difficult to distinguish a high-frequency component caused by a plant part from other noise components. As a result, when noise reduction (NR) processing is performed, there is a problem that an image of a plant part is blurred.

  Further, for example, it is conceivable to reduce the noise reduction intensity of the plant part by determining information indicating that the vicinity of the target pixel is a specific color, for example, information indicating that it is green, using the technique of Patent Document 1. However, when an artifact that is not a plant (for example, an automobile or a signboard) is green, there is a problem that noise included in the image of the artifact cannot be reduced.

  Therefore, the present invention has been made in view of the above problems, and an object of the present invention is to appropriately perform noise reduction processing on a distant view plant in an image including a distant view subject image. It is an object of the present invention to provide a new and improved noise reduction apparatus and noise reduction method that can obtain a sense of resolution.

  In order to solve the above-described problem, according to an aspect of the present invention, based on the pixel value of the target pixel of the image data and the pixel value of the pixel near the target pixel, an image composed of the target pixel and the target pixel vicinity is There is provided a noise reduction apparatus including a subject image determination unit that determines whether or not a subject image of a plant and a smoothing unit that smoothes a target pixel according to the determination result.

  With this configuration, based on the pixel value of the target pixel of the image data and the pixel value of the pixel in the vicinity of the target pixel, it is determined whether the image formed of the target pixel and the vicinity of the target pixel is a subject image of a distant plant. The target pixel is smoothed according to the determination result.

  The subject image determination unit includes: a high-frequency component calculation unit that calculates a high-frequency component in a pixel of interest and a pixel in the vicinity of the pixel of interest based on a pixel value of the pixel of interest and a pixel value in the vicinity of the pixel of interest; An average value of hue values of neighboring pixels, a hue saturation average value calculating unit for calculating an average value of saturation values of the target pixel and pixels near the target pixel, and a saturation value of the target pixel and pixels near the target pixel Saturation variance value calculation unit for calculating the variance value of the image, and the target pixel and the vicinity of the target pixel based on the calculated high-frequency component, the average value of the hue value, the average value of the saturation value, and the variance value of the saturation value And a determination unit that determines whether or not the image formed from is a subject image of a distant plant.

  The smoothing unit is a low-pass filter, and when it is determined that the image composed of the target pixel and the vicinity of the target pixel is a subject image of a distant plant compared to a case where it is determined that the image is not a distant plant subject image. Widen the transmission frequency range.

  The subject image determination unit further includes an infrared component value calculation unit that calculates an infrared component value among the pixel value of the target pixel and the pixel values of pixels near the target pixel.

  The determination unit includes the calculated high-frequency component that is greater than or equal to a predetermined value in the target pixel and pixels in the vicinity of the target pixel, and the calculated average value of the hue value and the average value of the saturation value are each included in the predetermined range. When the calculated variance value of the saturation value is equal to or greater than a predetermined value, it is determined that the image composed of the target pixel and the vicinity of the target pixel is a subject image of a distant plant.

  In the determination unit, when the calculated infrared component value is equal to or greater than a predetermined value that changes in accordance with the luminance value of the target pixel, an image including the target pixel and the vicinity of the target pixel is a subject image of a distant plant. Is determined.

  In order to solve the above problem, according to another aspect of the present invention, an image composed of a pixel of interest and a neighborhood of the pixel of interest based on a pixel value of the pixel of interest of image data and a pixel value of a pixel near the pixel of interest. There is provided a noise reduction method including a step of determining whether or not is a subject image of a distant plant and a step of smoothing a target pixel according to the determination result.

  The step of determining whether or not the subject image of a distant plant is based on the pixel value of the pixel of interest and the pixel value of the pixel near the pixel of interest, and calculates a high frequency component in the pixel of interest and the pixel near the pixel of interest. Calculating a mean value of hue values of the target pixel and pixels near the target pixel, a step of calculating an average value of saturation values of the target pixel and pixels near the target pixel, and a saturation of the target pixel and pixels near the target pixel. An image composed of the target pixel and the vicinity of the target pixel based on the step of calculating the variance value of the degree value and the calculated high-frequency component, the average value of the hue value, the average value of the saturation value, and the variance value of the saturation value Determining whether or not is a subject image of a distant plant.

  According to the above, even when an object having the same color (for example, green) as the subject of a distant plant is included in the subject image, the plant can be determined. Since it can be determined that the image is a plant subject image, when performing noise reduction processing on a pixel having the same color as the plant subject (for example, green), the noise reduction processing can be performed so that the image is not blurred. .

  According to the present invention, an image including a subject image of a distant view plant can be appropriately subjected to noise reduction processing for a distant view plant, and a sense of resolution can be obtained.

  Exemplary embodiments of the present invention will be described below in detail with reference to the accompanying drawings. In addition, in this specification and drawing, about the component which has the substantially same function structure, duplication description is abbreviate | omitted by attaching | subjecting the same code | symbol.

(Configuration of the first embodiment)
First, the configuration of the imaging apparatus 100 according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram illustrating an imaging apparatus 100 according to the present embodiment.

  The imaging apparatus 100 is a digital still camera that can capture a still image, for example. The imaging apparatus 100 includes, for example, a lens block 102, an imaging element 110, a preprocessing unit 120, an image memory 142, an image signal processing unit 144, an overall signal processing unit / control unit 146, a lens driving unit 148, a JPEG processing unit 150, and an LCD 160. , A memory card 170, an operation unit 180, and the like. Note that the imaging apparatus 100 may be a video camera that can capture a moving image.

  Although not shown, the lens block 102 includes, for example, a zoom lens, a diaphragm, and a focus lens. The lens block 102 is an optical system that forms an image of external light information on the image sensor 110, and transmits light from the subject to the image sensor 110. The zoom lens is a lens that changes the angle of view by changing the focal length. The diaphragm is a mechanism that adjusts the amount of light transmitted. The focus lens moves in the optical axis direction to focus the subject image on the imaging surface of the image sensor 110. The focus lens is driven by a lens driving unit 148.

  The image sensor 110 is an example of a photoelectric conversion element, and includes a plurality of elements capable of performing photoelectric conversion that converts optical information that has passed through the lens block 102 and entered into an electrical signal. Each element generates an electrical signal corresponding to the amount of light received. As the image sensor 110, a charge coupled device (CCD) sensor, a complementary metal oxide semiconductor (CMOS) sensor, or the like can be applied.

  In addition, in order to control the exposure time of the image sensor 110, a mechanical shutter (not shown) can be applied so that the light is blocked during non-shooting and the light is applied only during shooting. Further, the present invention is not limited to this, and an electronic shutter may be applied. The operation of the mechanical shutter or the electronic shutter is performed by a switch of a shutter button (operation unit 180) connected to the overall signal processing unit / control unit 146.

  The image sensor 110 further includes a CDS / AMP unit and an A / D conversion unit. The CDS / AMP unit (correlated double sampling circuit / amplifier) removes low-frequency noise contained in the electrical signal output from the image sensor 110, and brings the electrical signal to an arbitrary level. Amplify. The A / D conversion unit digitally converts the electrical signal output from the CDS / AMP unit to generate a digital signal. The A / D conversion unit outputs the generated digital signal to the preprocessing unit 120.

  The pre-processing unit 120 performs processing on the digital signal output from the A / D conversion unit, and generates an image signal that enables image processing. The preprocessing unit 120 performs processes such as pixel defect correction, black level correction, shading correction, and AF evaluation value calculation of the image sensor 110, for example. The preprocessing unit 120 outputs the generated image signal to the image signal processing unit 144, for example. Further, the preprocessing unit 120 controls reading / writing of image data from / to the image memory 142.

  The image memory 142 is, for example, an SDRAM (synchronous DRAM), and temporarily stores image data of captured images. The image memory 142 has a storage capacity capable of storing image data of a plurality of images. Reading and writing of images to and from the image memory 142 is controlled by the preprocessing unit 120.

  The image signal processing unit 144 receives the video signal from the preprocessing unit 120 and converts it into a luminance signal and a color signal. The image signal processing unit 144 includes a noise reduction unit 130 described later, and performs noise reduction processing. Further, the image signal processing unit 144 generates an image signal subjected to image processing based on the WB control value, the γ value, the edge (contour) enhancement control value, and the like. The image signal processing unit 144 sends the generated video signal to the overall signal processing unit / control unit 146.

  The overall signal processing unit / control unit 146 functions as an arithmetic processing device and a control device according to a program, and controls processing of each component provided in the imaging device 100. For example, the overall signal processing unit / control unit 146 outputs a signal to the lens driving unit 148 to drive the focus lens of the lens block 102. The overall signal processing unit / control unit 146 controls each component of the imaging apparatus 100 based on a signal from the operation unit 180. In the present embodiment, the entire signal processing unit / control unit 146 is composed of only one, but it is composed of a plurality of CPUs such as a separate CPU for signal system instructions and operation system instructions. May be.

  Upon receiving the focus control start operation signal, the lens driving unit 148 generates a control signal for moving the focus lens in one direction, and outputs the generated control signal to the lens block 102. The lens driving unit 148 calculates the focus position of the focus lens based on the AF evaluation value calculated by the preprocessing unit 120. The AF evaluation value is calculated by the preprocessing unit 120 based on the luminance value of the image signal. The AF evaluation value is, for example, a contrast value of an image, and when the contrast value is maximized, it is determined that the subject image is in focus on the imaging surface 110A of the image sensor 110 (contrast detection method). Note that the focus control of this embodiment is not limited to the contrast detection method, and may be performed by another method.

  The JPEG processing unit 150 performs compression encoding processing on the video data from the entire signal processing unit / control unit 146 by a still image encoding method such as JPEG (Joint Photographic Experts Group) standard. Further, decompression decoding processing is performed on the encoded data of the still image supplied from the memory card 170.

  The LCD 160 receives image data from, for example, a VRAM and displays an image on the screen. The LCD 160 is provided in the main body of the imaging device 100, for example. The image displayed on the LCD 160 is, for example, an image before shooting (live view display) read from the VRAM, various setting screens of the imaging apparatus 100, an image recorded and recorded, and the like. In the present embodiment, the LCD 160 is used as the display unit, but the present invention is not limited to this example, and may be an organic EL display, for example.

  A VRAM (video RAM) is a memory for image display and has a plurality of channels. The VRAM can simultaneously input image data for image display from the image memory 142 and output image data to the LCD 160. The resolution and maximum number of colors of the LCD 160 depend on the capacity of the VRAM.

  The memory card 170 writes image data or reads recorded image data and setting information. The memory card 170 is a recording medium such as a magnetic disk or a semiconductor storage medium, and records captured image data. As long as it is a recording medium, it is not limited to the memory card 170 but may be an optical disc (CD, DVD, Blu-ray disc, etc.), a magneto-optical disc, or the like. The memory card 170 may be configured to be detachable from the imaging device 100.

  The operation unit 180 is, for example, an up / down / left / right key, a power switch, a mode dial, or a shutter button provided in the imaging apparatus 100. The operation unit 180 sends an operation signal to the overall signal processing unit / control unit 146 and the like based on an operation by the user. For example, the shutter button can be pressed halfway (S1 operation), fully pressed (S2 operation), and released by the user. When the shutter button is half-pressed, it outputs an operation signal for starting focus control, and when the half-press is released, the focus control ends. Further, when the shutter button is fully pressed, an operation signal for starting shooting is output.

  Note that a series of processing in the imaging apparatus 100 may be processed by hardware or may be realized by software processing by a program on a computer.

(Configuration of noise reduction unit 130)
Next, the configuration of the noise reduction unit 130 according to the present embodiment will be described with reference to FIG. FIG. 2 is a block diagram illustrating the noise reduction unit 130 according to the present embodiment.

  The noise reduction unit 130 is an example of a noise reduction device, and is provided in the image signal processing unit 144, for example. The noise reduction unit 130 includes, for example, an input unit 131, a subject image determination unit 132, a filter processing selection unit 134, a filter processing unit 135, an output unit 136, and the like.

  The input unit 131 receives image data composed of a luminance-color difference signal after the image data output from the image sensor 110 is converted into a luminance signal and a color signal. The input image data is sent to the subject image determination unit 132.

  The subject image determination unit 132 determines whether or not the target pixel is the target image from the pixel value of the target pixel of the image data and the pixel values of the pixels near the target pixel. Here, the target subject image is a subject image of a plant such as a distant tree in the present embodiment. When a plant such as a tree is included in a distant view, the leaves of the tree (green blue leaves, autumn leaves of yellow, red, etc.) and flowers (sakura flowers, etc.) are very fine compared to the entire subject image. By determining whether or not the subject image is a distant plant, a portion determined to be a plant can be distinguished from normal noise.

  The subject image determination unit 132 includes, for example, a feature calculation unit 137, a determination unit 138, and the like. The feature calculation unit 137 calculates a reference value from which it can be determined that the target pixel is the subject image of interest. In the present embodiment, the subject image that is the object is a distant plant. The reference values for determining that the pixel of interest is a subject image of a distant plant are (1) high frequency component, (2) average value of hue value, average value of saturation value, and (3) saturation value. Variance value. The feature calculation unit 137 includes, for example, a high frequency component calculation unit 152, a hue saturation average value calculation unit 154, a saturation variance value calculation unit 156, and the like.

  The high frequency component calculation unit 152 calculates high frequency components in the pixel of interest and the pixels near the pixel of interest based on the pixel value of the pixel of interest and the pixel values of the pixels near the pixel of interest. The hue saturation average value calculation unit 154 calculates the average value of the hue values of the target pixel and the pixels near the target pixel based on the hue values of the target pixel and the pixels near the target pixel. Further, the hue saturation average value calculation unit 154 calculates the average value of the saturation values of the target pixel and the pixels near the target pixel based on the saturation values of the target pixel and the pixels near the target pixel. The saturation variance calculation unit 156 calculates the variance of the saturation values of the pixel of interest and the pixels near the pixel of interest based on the saturation values of the pixel of interest and the pixels near the pixel of interest.

  The determination unit 138 sets each value (for example, the calculated high frequency component, the average value of the hue value, the average value of the saturation value, and the variance value of the saturation value) that is the determination criterion calculated by the feature calculation unit 137. Based on this, it is determined whether or not the target pixel is a subject image (a subject image of a distant plant). For example, the determination unit 138 includes the calculated high-frequency component at a predetermined value or more in the target pixel and the pixels near the target pixel, and the calculated average value of the hue value and the average value of the saturation value are within the predetermined range, respectively. When the variance value of the calculated saturation value is greater than or equal to a predetermined value, it is determined that the image of the target pixel and the vicinity of the target pixel is a subject image of a distant plant.

  The filter processing selection unit 134 selects at least one filter from the plurality of filters recorded in the filter processing unit 135 based on the determination result of the determination unit 138. The selected filter differs depending on the image data (determination result of the determination unit 138). The filter processing selection unit 134 sends information about which filter should be selected to the filter processing unit 135.

  In the present embodiment, the filter processing unit 135 stores a plurality of types of filters in the storage unit. The filter processing unit 135 is an example of a smoothing unit, and the filter is, for example, a low-pass filter. Then, when it is determined that the image composed of the pixel of interest and the vicinity of the pixel of interest is a subject image of a distant plant, the frequency range to be transmitted is made wider than when it is determined that the image is not a subject image of a distant plant. The filter processing unit 135 receives information about which filter should be selected from the filter processing selection unit 134. Further, the filter processing unit 135 receives the image data from the input unit 131 and performs filter processing on the target pixel of the image data based on information on which filter should be selected. That is, the filter processing unit 135 smoothes the target pixel by the filter selected according to the image data by the filter processing selection unit 134, and reduces noise included in the image data. The filter processing unit 135 sends the filtered image data to the output unit 136.

  The output unit 136 outputs the filtered image data to the outside of the noise reduction unit 130. The output image data is subjected to, for example, color interpolation processing, white balance adjustment, gamma processing, and the like.

  The series of processing in the noise reduction unit 130 may be performed by hardware or may be realized by software processing using a program on a computer.

(Operation of the noise reduction unit 130)
Next, the operation of the noise reduction unit 130 according to the present embodiment will be described with reference to FIG. FIG. 3 is a flowchart showing the operation of the noise reduction unit 130 according to the present embodiment.

  First, image data including a luminance signal (Y) and color difference signals (Cb, Cr) is input to the input unit 131 of the noise reduction unit 130 (step S101). In addition, the image data of the YCbCr signal may be directly input to the noise reduction unit 130, or RAW data based on a signal output from an image sensor in which color filters are arranged in a Bayer array. Further, it may be a signal expressed in another color space such as an RGB signal in the RGB color space. When an RGB signal is input, a step of converting the signal into a luminance-color difference signal is added.

  Then, the feature calculation unit 137 calculates a reference value for determining that the target pixel is a distant plant from the pixel value of the target pixel of the input image data and the pixel values of the pixels near the target pixel.

As shown in FIG. 6, when the color difference signal Cb and the color difference signal Cr are expressed by orthogonal coordinates in a color space, the distance r from the origin represents the saturation, and the angle θ from the Cr axis represents the hue. FIG. 6 is a graph showing the color space represented by the CbCr signal. Therefore, for example, the hue saturation average value calculation unit 154 and the saturation dispersion value calculation unit 156 perform polar coordinate conversion on the color difference signals (Cr, Cb) of the pixel of interest and the pixels near the pixel of interest (step S102). Thus, the saturation r and the hue θ when the color difference signal (Cr, Cb) of a certain pixel is (V _Cr , V _Cb ) are known.

で表され、
また、色相θは、

で表される。 For example, the saturation r is

Represented by
The hue θ is

It is represented by

Next, the hue saturation average value calculation unit 154 calculates the saturation values r of the target pixel and the pixels near the target pixel using the following formula based on the saturation values r of the target pixel and the pixels near the target pixel. An average value is calculated (step S103).

Further, the hue saturation average value calculation unit 154 calculates the average value of the hue values θ of the target pixel and the pixels near the target pixel using the following formula based on the hue values θ of the target pixel and the pixels near the target pixel. Calculate (step S103).

Furthermore, based on the saturation value r of the pixel of interest and the pixels in the vicinity of the pixel of interest, the saturation dispersion value calculation unit 156 uses the following formula to calculate the variance of the saturation value r of the pixel of interest and the pixels in the vicinity of the pixel of interest. Is calculated (step S103).

  On the other hand, the high frequency component calculation unit 152 calculates high frequency components in the pixel of interest and the pixels near the pixel of interest based on the pixel value of the pixel of interest and the pixel values of the pixels near the pixel of interest. For example, the high frequency component calculation unit 152 detects the spatial frequency of the pixel of interest and the vicinity of the pixel of interest based on the pixel value of the pixel of interest and the pixel value of the pixel near the pixel of interest (step S104).

Specifically, two-dimensional DCT (discrete cosine transform) is performed on the image data to calculate DCT coefficients. For example, as shown in FIG. 4, a case where image data is divided into 5 × 5 pixel blocks will be described. FIG. 4 is an explanatory diagram showing an arrangement of pixels of image data. FIG. 4 shows a 5 × 5 pixel block extracted, and A in FIG. 4 represents a target pixel. If the pixel value of each pixel is f (x, y), the DCT coefficient F (u, v) for each block is represented by the following equation.

Thus, the DCT coefficients F (u, v) are arranged as shown in FIG. FIG. 5 is an explanatory diagram showing the arrangement of DCT coefficients. The upper left corner of each block represents a direct current (DC) component. Then, in order to determine whether or not the high frequency component is included in the target pixel and the pixels near the target pixel by a predetermined value or more, the coefficient k _fH is calculated by the following mathematical formula. Coefficient _{k fH} is the DCT coefficient of C1~C7 of 5, a root-sum-square of the DCT coefficients of D1 to D9.

  Next, the determination unit 138 uses each value (for example, the calculated high frequency component, the average value of the hue value, the average value of the saturation value, and the variance of the saturation value) that is the determination criterion calculated by the feature calculation unit 137. Based on (value), it is determined whether or not the target pixel is a subject image (a subject image of a distant plant) as a target (step S105).

を満たすか否かを判定する。 First, as condition 1, it is determined whether or not the coefficient k _fH is equal to or greater than a predetermined value. When the coefficient k _fH is relatively large, there are many high frequency components. This is because, if there are many high-frequency components, there is a possibility that distant plants are included in the image. For example,

It is determined whether or not the above is satisfied.

かつ

を満たすか否かを判定する。 Further, as condition 2, it is determined whether or not the average value of the saturation value r and the hue value θ is within a predetermined range. If the average value of the saturation value r and the hue value θ is close to, for example, the color of the leaves of the plant (green of green leaves, yellow or red of autumn leaves) or the color of flowers (light pink of cherry blossoms) It is because there is a possibility of being included. For example, to determine if it is a green area,

And

It is determined whether or not the above is satisfied.

を満たすか否かを判定する。 Further, as condition 3, it is determined whether or not the variance value of the saturation value r is greater than or equal to a predetermined value. This is because if the variance value (variation) of the saturation value r is large, the color change is large, and there is a possibility that distant plants are included in the image. For example,

It is determined whether or not the above is satisfied.

  When all the conditions 1 to 3 described above are satisfied, the determination unit 138 determines that the target pixel is a subject image of a distant plant. On the other hand, when any one of the conditions 1 to 3 is not satisfied, the determination unit 138 determines that the target pixel is not a subject image of a distant plant. The numerical values shown in the above conditions 1 to 3 are examples of values applied when the image has an 8-bit (0 to 255) level and has a noise of about ISO 100 with a compact camera. Therefore, when the ISO sensitivity is another value, the numerical values shown in the conditions 1 to 3 are different values.

  Then, the filter processing selection unit 134 selects at least one of the plurality of filters in the filter processing unit 135 depending on whether the target pixel is a subject image of a distant plant (Step S106).

  Thereafter, the filter of the selected filter processing unit 135 performs a filter process on the target pixel of the image data (step S107).

  For example, when it is determined that the target pixel is a subject image of a distant plant, a relatively weak LPF (low pass filter) is selected. FIG. 8 shows an example of a relatively weak LPF. Accordingly, a relatively weak LPF process is performed on the target pixel of the subject image of the distant view plant, and an appropriate noise reduction process can be performed on the subject image of the distant view plant.

  For example, when it is determined that the target pixel is not a subject image of a distant plant, a relatively strong LPF is selected. FIG. 9 shows an example of a relatively strong LPF. Accordingly, relatively strong LPF processing is performed on the target pixel other than the subject image of the distant plant, and appropriate noise reduction processing can be performed on the subject image other than the subject image of the distant plant. Finally, the filtered image data is output to the outside of the noise reduction unit 130 (step S108).

は、上述したステップＳ１０３で算出した彩度値ｒの平均値及び色相値θの平均値である。図７は、ＣｂＣｒ信号で表された色空間を示すグラフである。図７（Ｂ）は、人工物の被写体像におけるＣｂＣｒ信号の分布の一例を示す。 Hereinafter, the effect of the noise reduction unit 130 of the present embodiment will be described. For example, a subject image of an artifact having a flat (smooth) surface has substantially uniform variations in the saturation value direction and the hue value direction as shown in FIG. For this reason, the distribution of the color difference signals is a substantially circular distribution. In addition, relatively high frequency components are reduced. It is shown in the figure

Is the average value of the saturation value r and the average value of the hue value θ calculated in step S103 described above. FIG. 7 is a graph showing the color space represented by the CbCr signal. FIG. 7B shows an example of the distribution of the CbCr signal in the subject image of the artifact.

は、上述したステップＳ１０３で算出した彩度値ｒの平均値及び色相値θの平均値である。 On the other hand, the subject image of a distant plant has a large change in brightness and color, so that the variation of the color difference signal is large as shown in FIG. 7A, and the variation is particularly large in the saturation value r direction. For this reason, the distribution of the color difference signal is slightly elliptical. Moreover, a relatively high frequency component increases. FIG. 7A shows an example of the distribution of CbCr signals in a subject image of a distant plant. It is shown in the figure

Is the average value of the saturation value r and the average value of the hue value θ calculated in step S103 described above.

  As described above, the present embodiment focuses on the difference between the color difference signal distribution of the object image of the artifact and the object image of the distant plant and the high-frequency component, and in the present embodiment, the distant plant is included in the image. It is determined whether or not it is included.

  Thereby, it is possible to relatively weaken the noise reduction process for the subject image of the distant plant in the image data, and it is possible to obtain an image with a sense of resolution without blurring the image of the plant part. . On the other hand, a subject image other than a plant having a color component similar to that of a plant can be subjected to normal processing without weakening the noise reduction processing, and noise can be reliably reduced. For example, it is possible to avoid the problem that noise is not removed for an artifact (such as an automobile or a signboard) having a color close to that of a plant.

(Second Embodiment)
Next, an imaging apparatus 200 according to the second embodiment of the present invention will be described. In the present embodiment, the image sensor 110 is provided with a visible light removal filter that transmits only infrared rays, whereby the amount of infrared rays in the subject image can be detected. Then, by referring to the amount of infrared rays, it is determined whether the subject image of a distant plant is included in the image.

  For example, the subject image determination unit 132 in the noise reduction unit 130 further includes an infrared component value calculation unit. The infrared component value calculation unit calculates an infrared component value among the pixel value of the target pixel and the pixel values of pixels near the target pixel. Then, in addition to the conditions 1 to 3 described in the first embodiment, the determination unit 138 determines a condition 4 indicating whether or not the calculated infrared component value is equal to or greater than a predetermined value. This is because the amount of infrared rays from a plant is higher than that of an artificial object, and if the infrared component value is high, a distant view of the plant may be included in the image. The predetermined value is determined so as to change in proportion to the luminance value Y of the target pixel. This is because the amount of infrared rays increases as the luminance increases.

  When all of the above conditions 1 to 4 are satisfied, the determination unit 138 determines that the target pixel is a subject image of a distant plant. On the other hand, when any one of the conditions 1 to 4 is not satisfied, the determination unit 138 determines that the target pixel is not a subject image of a distant plant.

  In this way, by considering Condition 4 in addition to Conditions 1 to 3 described in the first embodiment, it is possible to determine whether or not the subject image is a distant plant object with higher accuracy. In the image including the image, it is possible to appropriately perform noise reduction processing for a distant plant.

  The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to such examples. It is obvious that a person having ordinary knowledge in the technical field to which the present invention pertains can come up with various changes or modifications within the scope of the technical idea described in the claims. Of course, it is understood that these also belong to the technical scope of the present invention.

  For example, in the above-described embodiment, the noise reduction unit 130 is provided in the image signal processing unit 144 and the noise reduction process is performed on the image data including the luminance signal and the color difference signal. It is not limited. For example, the noise reduction unit may perform noise reduction processing on image data (RAW data) in which image data is output from the image sensor 110 and pixel defect correction is performed. At this time, the noise reduction unit is provided in the preprocessing unit 120, for example.

In the above embodiment, the AC component C1 to C7, there has been described a case where calculating the coefficient k _fH using D1 to D9, the present invention is not limited to this example. For example, the AC component B1 to B5, C1 to C7, may calculate the coefficient _{k fH} using D1 to D9.

  Moreover, although the case where image data is divided into 5 × 5 pixel blocks has been described in the above embodiment, the present invention is not limited to this example. For example, an 8 × 8 pixel block or a 10 × 10 pixel block is used. The size of blocks to be divided is arbitrary. Further, the filter processing for noise reduction is not limited to the above-described filter, and may be performed by other processing.

  Moreover, although the said embodiment demonstrated the case where the kind of low-pass filter applied in the filter process part 135 was changed according to whether a to-be-photographed object is a distant plant, this invention is not limited to this example. For example, the low-pass filter to be applied may be the same regardless of whether or not the subject image is a distant plant. At this time, when filtering the original image, the target pixel to be filtered and the target not to be filtered The pixel ratio is changed according to whether the subject image is a distant plant.

1 is a block diagram illustrating an imaging apparatus according to a first embodiment of the present invention. It is a block diagram which shows the noise reduction part which concerns on the same embodiment. It is a flowchart which shows operation | movement of the noise reduction part which concerns on the same embodiment. It is explanatory drawing which shows the arrangement | sequence of the pixel of image data. It is explanatory drawing which shows the arrangement | sequence of a DCT coefficient. It is a graph which shows the color space represented by the CbCr signal. It is a graph which shows the color space represented by the CbCr signal. It is explanatory drawing which shows an example of a low-pass filter (LPF). It is explanatory drawing which shows an example of a low-pass filter (LPF).

Explanation of symbols

DESCRIPTION OF SYMBOLS 100,200 Image pick-up apparatus 102 Lens block 110 Image pick-up element 120 Preprocessing part 130 Noise reduction part 131 Input part 132 Subject image determination part 134 Filter processing selection part 135 Filter processing part 136 Output part 137 Feature calculation part 138 Determination part 142 Image memory 144 Image signal processing unit 146 Overall signal processing unit / control unit 148 Lens driving unit 150 JPEG processing unit 152 High frequency component calculation unit 154 Hue saturation average value calculation unit 156 Saturation variance value calculation unit 160 LCD
170 Memory card 180 Operation unit

Claims (7)

A subject image for determining whether or not the image of the target pixel and the vicinity of the target pixel is a subject image of a distant plant based on the pixel value of the target pixel of the image data and the pixel values of the pixels near the target pixel A determination unit;
The pixel of interest have a smoothing unit for smoothing in accordance with a result of the determination,
The smoothing unit is a low-pass filter,
Noise that widens the frequency range to be transmitted when it is determined that the image of the pixel of interest and the vicinity of the pixel of interest is a subject image of a distant plant compared to a subject image of a distant plant Reduction device. The subject image determination unit
A high frequency component calculation unit that calculates a high frequency component in the pixel of interest and a pixel in the vicinity of the pixel of interest based on a pixel value of the pixel of interest and a pixel value of the pixel in the vicinity of the pixel of interest;
An average value of hue values of the pixel of interest and pixels near the pixel of interest, and a hue saturation average value calculation unit for calculating an average value of saturation values of the pixel of interest and pixels near the pixel of interest;
A saturation dispersion value calculation unit for calculating a dispersion value of saturation values of the pixel of interest and pixels near the pixel of interest;
Based on the calculated high-frequency component, the average value of the hue value, the average value of the saturation value, and the variance value of the saturation value, an image composed of the pixel of interest and the vicinity of the pixel of interest is a distant view of a plant. The noise reduction device according to claim 1, further comprising: a determination unit that determines whether the image is a subject image. The subject image determination unit
Further comprising an infrared component value calculation section for calculating an infrared component value of the pixel values and the pixel values of the pixels of the target pixel near the pixel of interest, the noise reducing device according to claim 1 or 2. The determination unit includes the calculated high-frequency component including a predetermined value or more in the target pixel and pixels in the vicinity of the target pixel, and the calculated average value of the hue value and the average value of the saturation value are respectively When it is included in a predetermined range and the calculated variance value of the saturation value is equal to or greater than a predetermined value, it is determined that an image composed of the target pixel and the vicinity of the target pixel is a subject image of a distant plant. Item 3. The noise reduction device according to Item 2 . The determination unit further includes a subject of a distant plant where the image of the target pixel and the vicinity of the target pixel is a distant object when the calculated infrared component value is equal to or greater than a predetermined value that changes according to the luminance value of the target pixel. The noise reduction device according to claim 3 , wherein the noise reduction device is determined to be an image. Determining whether or not the image of the target pixel and the vicinity of the target pixel is a subject image of a distant plant based on the pixel value of the target pixel of the image data and the pixel value of the pixel near the target pixel; ,
Smoothing the target pixel according to the result of the determination,
The smoothing step is performed by a low-pass filter,
Noise that widens the frequency range to be transmitted when it is determined that the image of the pixel of interest and the vicinity of the pixel of interest is a subject image of a distant plant compared to a subject image of a distant plant Reduction method. The step of determining whether or not the subject image of the distant plant is:
Calculating a high frequency component in the pixel of interest and a pixel near the pixel of interest based on a pixel value of the pixel of interest and a pixel value of a pixel near the pixel of interest;
Calculating an average value of hue values of the target pixel and pixels near the target pixel, and an average value of saturation values of the target pixel and pixels near the target pixel;
Calculating a variance value of saturation values of the target pixel and pixels in the vicinity of the target pixel;
Based on the calculated high-frequency component, the average value of the hue value, the average value of the saturation value, and the variance value of the saturation value, an image composed of the pixel of interest and the vicinity of the pixel of interest is a distant view of a plant. The noise reduction method according to claim 6 , further comprising a step of determining whether or not the image is a subject image.

Images