JP2018106763A - Imaging apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To properly evaluate an image in which an unnecessary object is overlapped on a face.SOLUTION: An imaging apparatus 1 comprises: face detection means 11 which detects a face of a subject from an inputted image; area setting means 11 which sets a plurality of small areas in an area of the face; feature amount calculation means 11 which calculates feature amounts of the plurality of small areas; and image evaluation means 11 which evaluates the image on the basis of difference between the feature amounts of the small areas on a face left side and the feature amounts of the small areas on a face right side among the plurality of small areas.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an imaging apparatus.

  There is known a camera that selects a best shot candidate image from a plurality of continuously shot images (see Patent Document 1). With this camera, the above multiple images are evaluated for multiple evaluation factors, such as the presence or absence of unwanted facial parts, the presence or absence of eye meditation, and blurring of the face area, and the best shot image is selected based on the evaluation results. It is.

JP 2012-195949 A

  In the above-described conventional technology, the evaluation is lowered so that an image in which an unnecessary object overlaps the face of the subject is not selected as the best shot candidate image. The evaluation of whether or not an unnecessary object is superimposed on the face can be determined by detecting that a part of face parts (eyes, nose, mouth, etc.) are not detected as a result of the face detection process. However, for images where facial parts are not occluded even if unwanted objects overlap the face, it is not possible to determine whether the unwanted parts are superimposed on the face based on the detection results of the facial parts, and appropriate evaluation can be performed. There is a risk of not.

  An imaging apparatus according to an aspect of the present invention includes a face detection unit that detects a face of a subject from a plurality of input images, and a first region and a second region that include the face region detected by the face detection unit. An evaluation unit that evaluates each of the plurality of images based on the characteristics.

  According to the present invention, it is possible to appropriately evaluate a plurality of images.

It is a block diagram explaining the principal part structure of the digital camera by one embodiment of this invention. It is a figure explaining the acquisition timing of the image in pre capture photography mode. It is a flowchart explaining the flow of the process in pre capture photography mode. It is a figure which shows an example of a picked-up image typically. It is a flowchart explaining the flow of a face detection reliability evaluation process. It is a figure explaining the several small area | region which divides | segments a face area. It is a figure which shows the state which set the some small area | region which divides | segments a face area | region about the image with which the unnecessary object has overlapped on the face.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. FIG. 1 is a block diagram for explaining a main configuration of a digital camera 1 according to an embodiment of the present invention. The digital camera 1 is controlled by the main CPU 11.

  The photographing lens 21 forms a subject image on the imaging surface of the imaging element 22. The image pickup device 22 is configured by a CCD image sensor or a CMOS image sensor, picks up a subject image on the image pickup surface, and outputs an image pickup signal to the image pickup circuit 23. The imaging circuit 23 performs analog processing (such as gain control) on the photoelectric conversion signal output from the imaging element 22, and converts the analog imaging signal into digital data by a built-in A / D conversion circuit.

  The main CPU 11 inputs a signal output from each block, performs a predetermined calculation, and outputs a control signal based on the calculation result to each block. The image processing circuit 12 is configured as an ASIC, for example, and performs image processing on the digital image signal input from the imaging circuit 23. The image processing includes, for example, contour enhancement, color temperature adjustment (white balance adjustment) processing, and format conversion processing for an image signal.

  The image compression circuit 13 performs image compression processing at a predetermined compression ratio on the image signal processed by the image processing circuit 12 by, for example, the JPEG method. The display image creation circuit 14 generates a display signal for displaying the captured image on the liquid crystal monitor 19.

  The liquid crystal monitor 19 includes a liquid crystal panel, and displays an image, an operation menu screen, and the like based on a display signal input from the display image creation circuit 14. The video output circuit 20 generates and outputs a video signal for displaying an image, an operation menu screen, or the like on an external display device based on the display signal input from the display image creation circuit 14.

  The buffer memory 15 temporarily stores data before image processing, after image processing, and during image processing, and also stores an image file before being recorded on the recording medium 30 and an image file read from the recording medium 30. Used to remember. In the present embodiment, it is also used when a pre-capture image acquired at a predetermined frame rate by the image sensor 22 is temporarily stored before a shooting instruction (before the release button is fully pressed). The pre-capture image will be described later.

  The flash memory 16 stores programs executed by the main CPU 11, data necessary for processing performed by the main CPU 11, and the like. The contents of the program and data stored in the flash memory 16 can be added or changed by an instruction from the main CPU 11.

  The card interface (I / F) 17 has a connector (not shown), and a recording medium 30 such as a memory card is connected to the connector. The card interface 17 writes data to the connected recording medium 30 and reads data from the recording medium 30 in accordance with an instruction from the main CPU 11. The recording medium 30 is configured by a memory card incorporating a semiconductor memory, a hard disk drive, or the like.

  The operation member 18 includes various buttons and switches of the digital camera 1 and outputs an operation signal according to the operation content of each operation member, such as a switching operation of a mode change switch, to the main CPU 11. The half-press switch 18a and the full-press switch 18b each output an ON signal to the main CPU 11 in conjunction with a pressing operation of a release button (not shown). The ON signal (half-press operation signal) from the half-push switch 18a is output when the release button is pushed down to about half of the normal stroke, and the output is released when the half-stroke push-down operation is released. The ON signal (full push operation signal) from the full push switch 18b is output when the release button is pushed down to the normal stroke, and the output is released when the normal stroke push down operation is released. The half-press operation signal instructs the main CPU 11 to start shooting preparation. The full-press operation signal instructs the main CPU 11 to start acquiring a recording image.

<Shooting mode>
The digital camera 1 has a normal shooting mode and a pre-capture shooting mode. In the normal shooting mode, a shot image is acquired frame by frame in accordance with the full-press operation signal and recorded on the recording medium 30. In pre-capture shooting mode, when a half-press operation signal is received, a still image at a high shutter speed (for example, faster than 1/125 second) is shot continuously at 120 frames / second (120 fps), and multiple frames are shot continuously. Acquire a captured image. When the full-press operation signal is received, predetermined frame images before and after receiving the full-press operation signal are recorded on the recording medium 30, respectively. Each photographing mode is configured to be switchable according to an operation signal from the operation member 18.

<Playback mode>
The digital camera 1 in the reproduction mode reproduces and displays on the liquid crystal monitor 19 the image recorded in each of the above photographing modes for each frame or for each predetermined number of frames.

<Pre-capture shooting>
FIG. 2 is a diagram for explaining image acquisition timing in the pre-capture shooting mode. In FIG. 2, when a half-press operation signal is input at time t0, the main CPU 11 starts a release standby process. In the release standby process, for example, subject images are continuously captured at a frame rate of 120 frames / second (120 fps), and the acquired image data is stored in the buffer memory 15 sequentially.

  A sufficient memory capacity of the buffer memory 15 used for pre-capture photography is secured in advance. When the number of frame images stored in the buffer memory 15 after the time t0 reaches a predetermined number (for example, A), the main CPU 11 overwrites and erases the oldest frame images in order. Thereby, the memory capacity of the buffer memory 15 used for pre-capture photography can be limited.

  When the full press operation signal is input at time t1, the main CPU 11 starts the release process. In the release process, (A + B) frame images obtained by combining A frame images captured before time t1 and B frame images captured after time t1 are used as recording candidate images on the recording medium 30. To do.

  Next, the flow of processing in the pre-capture shooting mode will be described using the flowchart shown in FIG. When the half-press operation signal is input, the main CPU 11 starts the processing shown in FIG.

  In step S1 of FIG. 3, the main CPU 11 starts the release standby process described above, and proceeds to step S2.

  In step S2, the main CPU 11 determines whether or not a full-press operation signal has been input. The CPU 11 repeats the determination process in step S2 until a full-press operation signal is input. When the full-press operation signal is input, the CPU 11 proceeds to step S3.

  In step S3, the main CPU 11 performs the above-described release process, sets (A + B) frame images as recording candidate images, and proceeds to step S4.

  In step S4, the main CPU 11 performs an image evaluation process for evaluating each of the (A + B) frame images as recording candidate images. In this image evaluation process, the main CPU 11 evaluates each of (A + B) frame images with respect to a plurality of evaluation items such as face detection reliability, presence / absence of eye meditation, and blurring of the face area. Then, the main CPU 11 obtains a comprehensive evaluation for each frame image (A + B) frame images based on these evaluation results, and proceeds to step S5.

  In step S5, based on the evaluation result in step S4, the main CPU 11 performs the best shot on C frame images (C <(A + B)) having a high overall evaluation among (A + B) frame images. Select as a candidate image. The main CPU 11 records the selected C frame images in the recording medium 30 in association with each other. Further, the main CPU 11 selects one frame image having the highest overall evaluation among the C frame images (best shot candidate images) as the best shot image, and ends the process of FIG.

  The present embodiment has a feature in the above-described image evaluation processing, particularly in the face detection reliability evaluation processing. Therefore, the following description will be focused on this point.

<Face detection reliability evaluation process>
First, an outline of the face detection reliability evaluation process will be described with reference to FIG. FIG. 4 schematically illustrates an example of an image continuously shot when a person who is a subject is waving his hand 150 in front of the face 100. In the image shown in FIG. 4A, the hand 150 overlaps the person's face 100, and the person's eyes are visible, but the nose and mouth are shielded. In the image shown in FIG. 4B, the hand 150 overlaps a part of the person's face 100, but the eyes, nose and mouth of the person are not shielded. In the image shown in FIG. 4C, the hand 150 does not overlap the face 100 of the person. For example, unlike when a person poses with his hand 150 on the face, when the hand 150 is moving, the person 150 unintentionally overlaps the face and the hand 150 is an unnecessary object. Conceivable. Therefore, among the images shown in FIGS. 4A to 4C, the image shown in FIG. 4C in which the unnecessary hand 150 does not overlap the human face 100 is selected as the best shot candidate image. It is desirable to be done. Therefore, in the images shown in FIGS. 4A and 4B, the face detection reliability is evaluated relatively low, and in the image shown in FIG. 4C, the face detection reliability is evaluated relatively high. It is desirable.

  When face detection processing is performed on the image shown in FIG. 4A, the skin color and eyes of the face 100 are detected, but the nose and mouth are not detected. Therefore, it can be determined from the detection result of the face parts that an unnecessary object is overlapped with the face 100, and the evaluation of the face detection reliability can be lowered.

  On the other hand, when the face detection process is performed on the image shown in FIG. 4B, the eyes, nose, and mouth are detected even though the hand 150 overlaps the chin portion of the face 100. Similarly, eyes, nose and mouth are also detected when face detection processing is performed on the image shown in FIG. Therefore, the image shown in FIG. 4B cannot be distinguished from the image shown in FIG. 4C only by the detection result of the face parts, and there is an unnecessary object on the face 100 in the image shown in FIG. 4B. I cannot judge that they overlap. Therefore, also in the image shown in FIG. 4B, the evaluation of the face detection reliability becomes high as in the image shown in FIG.

  Therefore, in this embodiment, as shown in FIG. 4B, the evaluation is low for an image in which the eyes, nose, and mouth are not shielded but an unnecessary object overlaps the face 100, as shown in FIG. In addition, the evaluation of the face detection reliability is performed so that the evaluation is high for an image in which an unnecessary object does not overlap the face 100.

  Details of the face detection reliability evaluation process will be described below. FIG. 5 is a flowchart for explaining the flow of the face detection reliability evaluation process. When the release process is completed and (A + B) frame images as recording candidate images are input to the buffer memory 15, the main CPU 11 starts the process of FIG.

  In step S11 of FIG. 5, the main CPU 11 performs face detection processing for detecting the face of the subject in each of the (A + B) frame images that are the recording candidate images. A known method may be used for the face detection process.

  In step S12, as shown in FIG. 6, the main CPU 11 sets seven small rectangular areas 201 to 207 that divide the area of the face 100 detected by the face detection process. These seven small areas 201 to 207 are set to positions and sizes including the outline of a part of the face.

  Specifically, the main CPU 11 sets the left eye small region 201 so as to include the contour of the left eye 101, and sets the right eye small region 202 so as to include the contour of the right eye 102. The left-eye small area 201 and the right-eye small area 202 are set so as to divide the eye and the surrounding area into left and right from the center position of the left eye 101 and the right eye 102.

  Further, the main CPU 11 sets the left cheek small region 203 so as to include the left cheek contour, and sets the right cheek small region 204 so as to include the right cheek contour. The left cheek small area 203 is set below the left eye small area 201. The right cheek small area 204 is set below the right eye small area 202. The left cheek small area 203 and the right cheek small area 204 are set so as to divide the nose and cheek areas from the center position of the nose 103 to the left and right.

  Further, the main CPU 11 sets the left jaw small region 205 so as to include the contour of the left jaw, sets the right jaw small region 206 so as to include the contour of the right jaw, and sets the mouth small size so as to include the contour of the mouth 104. An area 207 is set. The left jaw small area 205 is set below the left cheek small area 203 and on the left side of the mouth small area 207. The right jaw small area 206 is set on the right side of the mouth small area 207.

  In the present embodiment, the contour (edge) itself of the face 100 is not detected by the face detection process. However, the position of the standard contour of the face 100 can be estimated from the positional relationship between the left eye 101, the right eye 102, and the mouth 104. Based on this, the main CPU 11 determines the positions and sizes of the seven small areas 201 to 207 based on the positional relationship between the left eye 101, the right eye 102, and the mouth 104 obtained as a result of the face detection process.

  In step S <b> 13 of FIG. 5, the main CPU 11 calculates the sharpness in each of the set seven small regions 201 to 207. Note that the sharpness of each small region may be calculated by a known method, for example, using a secondary differentiation process.

  FIG. 7 is a diagram showing a state in which the seven small areas 201 to 207 are set in the area of the face 100 of the image shown in FIG. In FIG. 7, in the face 100, the eyes, nose, and mouth are not shielded, but the hand 150 overlaps from the left jaw to the mouth, so a part of the contour of the left jaw is shielded. Therefore, in FIG. 7, the sharpness of the left jaw subregion 205 is lower than the sharpness of the right jaw subregion 206 where the contour of the jaw portion is not shielded. In particular, when the image is taken while the hand 150 is moved, the hand 150 is blurred and the sharpness of the small left jaw region 205 is lower than the sharpness of the small right jaw region 206.

  For example, as shown in FIG. 6, in the image in which the hand 150 does not overlap the face 100, it is assumed that the ratio between the sharpness of the left jaw subregion 205 and the sharpness of the right jaw subregion 206 is 10: 9. To do. On the other hand, as shown in FIG. 7, in the image in which the hand 150 overlaps the left jaw part, for example, the ratio of the sharpness of the left jaw small area 205 to the sharpness of the right jaw small area 206 is 5: 9. For example, the sharpness of the left jaw subregion 205 is lower than that of the image shown in FIG. That is, when the image shown in FIG. 6 is compared with the image shown in FIG. 7, the degree of difference between the sharpness of the left jaw subregion 205 and the sharpness of the right jaw subregion 206 is different.

  Therefore, the image of the evaluation target is compared by comparing the image where the face 100 is not occluded with the image of the evaluation target with respect to the degree of difference between the sharpness of the left jaw small region 205 and the sharpness of the right jaw small region 206. It can be determined whether or not an unnecessary object overlaps the left jaw or the right jaw. That is, when the degree in the image to be evaluated is significantly different from the degree in the image where the face 100 is not shielded, it can be determined that an unnecessary object overlaps the left jaw or the right jaw in the image to be evaluated. . On the other hand, when the above-mentioned degree in the image where the face 100 is not shielded is substantially equal to the above-mentioned degree in the evaluation target image, it can be determined that no unnecessary object overlaps the left jaw or the right jaw in the evaluation target image.

  Further, it can be considered that the left eye small area 201 and the right eye small area 202 and the left cheek small area 203 and the right cheek small area 204 can be similarly determined. That is, it is considered that it can be determined whether or not an unnecessary object overlaps the face 100 based on the difference between the sharpness of the small area on the left side of the face and the sharpness of the small area on the right side of the face.

  Furthermore, not only the difference in sharpness between the left and right subregions, but also the difference in sharpness between the left cheek subregion 203 and the left jaw subregion 205, and between the right cheek subregion 204 and the right jaw subregion 206 are compared. However, it can be judged in the same way. Further, if the difference between the average value of the sharpness of the small left eye area 201 and the small right eye area 202 is compared with the sharpness of the small mouth area 207, it can be determined whether or not an unnecessary object overlaps the eyes or mouth. It is done.

  In the present embodiment, the evaluation process of the face detection reliability is performed based on the above points. In step S14 of FIG. 5, the main CPU 11 determines, from the (A + B) frame images that are recording candidate images, one frame image that is predicted not to be covered by the face 100 as a reference image. . At this time, the main CPU 11 determines a reference image based on the sharpness calculated in each of the seven small regions 201 to 207 of each frame image in step S13.

  For example, the main CPU 11 obtains the maximum value of the sharpness in each of the seven small areas 201 to 207 in (A + B) frame images. The main CPU 11 calculates the difference between the sharpness of each frame image and the maximum value of the sharpness, and the face 100 is shielded from the frame image in which the difference in the seven small regions 201 to 207 is the smallest overall. It is determined that there is no image and determined as a reference image.

In step S15, the main CPU 11 sequentially sets (A + B) frame images as evaluation target images, and evaluates the face detection reliability. The main CPU 11 compares the reference image and the evaluation target image for the degree of difference in sharpness in the following first to sixth groups, and evaluates the face detection reliability in the evaluation target image based on the comparison result. .
(First set) The degree of difference between the sharpness of the left eye small area 201 and the sharpness of the right eye small area 202 (second set) The difference between the sharpness of the left cheek small area 203 and the sharpness of the right cheek small area 204 Of the left jaw subregion 205 and the sharpness of the right jaw subregion 206 (fourth set) The sharpness of the left cheek subregion 203 and the sharpness of the left jaw subregion 205 Degree of difference between degrees (5th set) Degree of difference between sharpness of right cheek small area 204 and sharpness of right jaw small area 206 (6th set) Sharpness of left eye small area 201 and right eye small area 202 Of the difference between the average value of the sharpness and the sharpness of the small mouth area 207

  As a method for comparing the reference image and the evaluation target image with respect to the degree of difference in the sharpness, for example, a method described below may be used. In the following description, an example will be described in which the degree of difference in sharpness in the third set (the set of the left jaw subregion 205 and the right jaw subregion 206) is compared. The main CPU 11 normalizes the sharpness of the left jaw subregion 205 in the evaluation target image with the sharpness of the left jaw subregion 205 in the reference image. Further, the main CPU 11 normalizes the sharpness of the right jaw small region 206 in the evaluation target image with the sharpness of the right jaw small region 206 in the reference image. Here, the sharpness of the left jaw subregion 205 after normalization in the evaluation target image is expressed as α, and the sharpness of the right jaw subregion 206 is expressed as β.

  When the ratio of α and β is within a predetermined range from the reference value, the main CPU 11 determines that the difference between the degree in the evaluation target image and the degree in the reference image is within the predetermined range. In this case, the main CPU 11 determines that an unnecessary object does not overlap the left jaw subregion 205 or the right jaw subregion 206.

  On the other hand, when the ratio of α and β differs from the reference value by a predetermined value or more, the main CPU 11 determines that the degree in the evaluation target image is different from the degree in the reference image by a predetermined value or more. In this case, the main CPU 11 determines that an unnecessary object overlaps the left jaw subregion 205 or the right jaw subregion 206.

  In this way, the main CPU 11 compares the reference image with the evaluation target image for the degree of difference in sharpness in the first to sixth groups, and determines whether or not unnecessary objects overlap each small region. to decide. For any one of the first to sixth sets, the main CPU 11 determines the degree of face detection reliability in the evaluation target image when the degree in the evaluation target image differs from the above level in the reference image by a predetermined amount or more. The evaluation is relatively low in the (A + B) frame images. That is, the main CPU 11 relatively lowers the evaluation of the face detection reliability when it is determined that any one of the first to sixth sets overlaps with an unnecessary object.

  On the other hand, if the difference between the degree in the evaluation target image and the degree in the reference image is within a predetermined range in all of the first to sixth sets, the main CPU 11 (A + B) ) Relatively high in one frame image. That is, the main CPU 11 relatively increases the evaluation of the face detection reliability when it is determined that unnecessary objects do not overlap in all of the first to sixth groups.

  As described above, when it is determined that any one of the first group to the sixth group overlaps the unnecessary object, the main CPU 11 determines the unnecessary object in all the first group to the sixth group. The evaluation of the face detection reliability in the evaluation target image is made lower than when it is determined that the two do not overlap.

  As a result of the above processing, not only the eye, nose and mouth are shielded as shown in FIG. 4 (a), but also the eyes, nose and mouth are shielded as shown in FIG. 4 (b). The evaluation of the face detection reliability can be relatively lowered even for an image in which an unnecessary object overlaps the face 100. On the other hand, as shown in FIG. 4C, the evaluation of the face detection reliability can be relatively increased for an image in which an unnecessary object does not overlap the face 100.

  In step S5 of FIG. 3 described above, the main CPU 11 selects C best shot candidate images from (A + B) frame images based on the evaluation result including the evaluation of the face detection reliability. At this time, since the evaluation of the face detection reliability is relatively low with respect to an image in which the eyes, nose and mouth are not shielded but an unnecessary object is superimposed on the face 100, such an image is the best shot candidate image. Can be prevented from being selected. On the other hand, since the evaluation of the face detection reliability is relatively high for an image in which an unnecessary object does not overlap the face 100, it can be selected as a best shot candidate image.

According to the embodiment described above, the following operational effects can be obtained.
(1) In the digital camera 1, the main CPU 11 detects the subject's face from a plurality of input images, sets a plurality of small areas 201 to 207 in the face area, and each of the plurality of small areas 201 to 207. In, the sharpness is calculated. The main CPU 11 determines a reference image from the plurality of images. The main CPU 11 determines the difference between the sharpness of the small area on the left side of the plurality of small areas 201 to 207 and the sharpness of the small area on the right side of the face between the reference image and the other plurality of images. The plurality of images are each evaluated based on the comparison result. The main CPU 11 selects one or more images from the plurality of images based on the evaluation result. By doing so, the digital camera 1 can appropriately evaluate an image in which an unnecessary object is superimposed on the face, and can prevent such an image from being selected as a best shot candidate.

(2) When the degree of the evaluation target image among the plurality of images is different from the degree of the reference image by a predetermined amount or more, the main CPU 11 relatively evaluates the evaluation target image within the plurality of images. To lower. Thereby, the evaluation of an image in which an unnecessary object is superimposed on the face can be relatively lowered in the plurality of images.

(3) The plurality of small regions 201 to 207 are set to positions and sizes including a partial outline of the face. Thereby, based on the sharpness calculated in each of the plurality of small regions 201 to 207, it can be determined whether or not an unnecessary object is overlapping the face.

(4) The main CPU 11 compares the degree of difference between the sharpness of the left cheek subregion 203 and the sharpness of the right cheek subregion 204 between the reference image and the other plurality of images. Thereby, it can be determined whether or not an unnecessary object overlaps the left cheek or the right cheek.

(5) The main CPU 11 compares the degree of difference between the sharpness of the left jaw subregion 205 and the sharpness of the right jaw subregion 206 between the reference image and the other plurality of images. Thereby, it can be judged whether an unnecessary thing has overlapped with the left jaw or the right jaw.

(6) The main CPU 11 compares the degree of difference between the sharpness of the left-eye small area 201 and the sharpness of the right-eye small area 202 between the reference image and the other plurality of images. Thereby, it can be determined whether or not an unnecessary object overlaps the left eye or the right eye.

(7) The main CPU 11 compares the degree of difference between the sharpness of the left cheek subregion 203 and the sharpness of the left jaw subregion 205 between the reference image and the other plurality of images. Thereby, it can be determined whether or not an unnecessary object overlaps the left cheek or the left jaw.

(8) The main CPU 11 compares the degree of difference between the sharpness of the right cheek subregion 204 and the sharpness of the right jaw subregion 206 between the reference image and the other plurality of images. Thereby, it can be determined whether or not an unnecessary object overlaps the right cheek or the right jaw.

(9) The main CPU 11 compares the degree of difference between the sharpness of the small left area 201 and the small right eye area 202 and the sharpness of the small mouth area 207 between the reference image and the other plurality of images. Thereby, it can be determined whether or not unnecessary objects overlap the eyes or mouth.

(Modification 1)
In the above-described embodiment, as shown in FIG. 6, the example in which the face area is divided into seven small areas has been described. However, the division method such as the number of divisions of the face area and the shape of the small area is not limited to this. For example, the left jaw subregion 205 and the right jaw subregion 206 may be set so that the chin region including the mouth is divided into left and right from the center position of the mouth 104 without setting the mouth subregion 207. Good.

  In the above-described embodiment, the example in which the small area is set to divide the face area vertically and horizontally has been described. However, the small area may be set to divide the face area only to the left and right. You may set a small area | region so that an area | region may be divided | segmented only into upper and lower sides.

(Modification 2)
In the embodiment described above, the reference image and the evaluation target image are compared for the degree of difference in sharpness in the first to sixth sets of small regions, and the face detection reliability in the evaluation target image is based on the comparison result. The example which evaluates a degree was demonstrated. However, it is not always necessary to compare the degree of sharpness difference for all of the first to sixth groups.

  For example, the reference image and the evaluation target image are compared for the degree of difference in sharpness in the first to third sets of small regions, and the face detection reliability in the evaluation target image is evaluated based on the comparison result. It may be. Even in this case, it can be evaluated by judging whether or not an unnecessary object is superimposed on the face 100.

(Modification 3)
In the embodiment described above, an example has been described in which the sharpness is calculated as the feature amount in each small region set as the face region, and the face detection reliability is evaluated based on the sharpness. However, in each small region, another feature amount may be calculated instead of the sharpness, and the face detection reliability may be evaluated based on the feature amount. For example, in each small region, the smoothness of the face outline may be calculated, and the face detection reliability may be evaluated based on the smoothness of the face outline. In addition, for example, pattern matching of face contour lines may be performed in each small region, and the face detection reliability may be evaluated based on the result. In this case, an image in which an unnecessary object (such as a hand) does not overlap the face may be determined as the reference image, and pattern matching of the face contour line may be performed based on the reference image for images other than the reference image.

  Although various embodiments and modifications have been described above, the present invention is not limited to these contents. Other embodiments conceivable within the scope of the technical idea of the present invention are also included in the scope of the present invention.

DESCRIPTION OF SYMBOLS 1 ... Digital camera, 11 ... Main CPU, 15 ... Buffer memory, 18 ... Operation member, 18a ... Half press switch, 18b ... Full press switch, 21 ... Shooting lens, 22 ... Imaging element, 30 ... Recording medium

Claims (7)

A face detection unit for detecting the face of a subject from a plurality of input images;
An evaluation unit that evaluates each image of the plurality of images based on features of the first region and the second region including the face region detected by the face detection unit;
An imaging apparatus comprising: The imaging device according to claim 1,
A reference image determination unit for determining a reference image from the plurality of images;
The evaluation unit is based on the feature of the first region and the feature of the second region in the reference image, and the feature of the first region and the feature of the second region in an image other than the reference image. An imaging device that performs the evaluation. The imaging device according to claim 2,
The evaluation unit compares the reference image with another image other than the reference image for the degree of difference between the feature of the first region and the feature of the second region, and based on the result of the comparison, An imaging device that evaluates an image. The imaging device according to claim 3.
The evaluation unit is configured to make the evaluation of an image relatively low in the plurality of images when the degree of the image other than the reference image is different from the degree of the reference image by a predetermined amount or more. In the imaging device according to any one of claims 1 to 4,
An imaging apparatus comprising: an area setting unit that sets the first area and the second area at a position and size including a partial outline of the face. In the imaging device according to any one of claims 1 to 5,
The imaging device, wherein the plurality of images are images taken continuously. In the imaging device according to any one of claims 1 to 6,
An imaging apparatus further comprising an image selection unit that selects one or more images from the plurality of images based on an evaluation result by the evaluation unit.

Images