JP2007110528A - Imaging apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To correctly detect hand shake even when an imaging device is exposed for each pixel or each line. <P>SOLUTION: An imaging device 20 is exposed for each pixel, and generates an image signal of a photographed image of one frame. The imaging element 20 continuously generates an image signal of a photographed image of two or more frames. An image feature extracting device 72 extracts a feature of a subject image. First and second angular velocity sensors 41 and 42 detect the angular velocity of a camera body. An angular velocity-motion vector converting device 61 applies time integration to the angular velocity in each of timings when the feature forms an image is exposed, consequently detecting shake to be generated between each of photographed images. An image shifter 24 shifts each of the photographed images in order to correct relative positional deviation, generated due to shake to be generated in each of the photographed images, between an initial photographed image and each of the photographed images. An image multiplexer 26 multiplexes each of the shifted photographed images to obtain a multiplexed image. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an imaging apparatus having an image blur correction function, and more particularly to a digital camera.

  Conventionally, in an imaging apparatus such as a digital camera, in order to correct camera shake that occurs during imaging, image blur that detects the vibration of the imaging apparatus and corrects the position of the imaging optical system according to the detected vibration amount Correction devices are known.

  Further, as a method for easily correcting image blur, a method in which correction is performed by image processing is known, and an example of this method is a method described in Patent Document 1. That is, the captured image is continuously captured for a plurality of frames, and the amount of image blur when each captured image is captured is detected, and each captured image is added after being shifted according to the amount of image blur. Two still images are generated. According to this method, by shortening the time (exposure time) during which each captured image is captured, image blurring that occurs during the capture of one frame of captured image can be minimized, and between each frame. The positional deviation is canceled by shifting each captured image, and thus image blur correction is appropriately performed.

In Patent Document 1, a CCD is used as an image sensor, and all pixels are exposed at the same timing for the same period, and the electric charge accumulated during the exposure is converted into an electric signal. Therefore, each pixel of each captured image has the same image blur amount. Therefore, when shifting each captured image, it is not necessary to consider the difference in the amount of image blur between the pixels in the same captured image.
Japanese Patent Laid-Open No. 2003-32540

  However, a CMOS sensor is known as an image sensor in addition to a CCD. In the CMOS sensor, a subject image is sequentially photoelectrically converted for each pixel to generate a one-frame captured image. That is, since the exposure timing of each pixel differs in one captured image, the amount of image blur of each pixel of the captured image varies from pixel to pixel. Therefore, in detecting a positional shift between a plurality of captured images, even if the method described in Patent Document 1 is applied as it is, there is a case where images cannot be combined appropriately. In addition, there is an example in which a CMOS sensor performs photoelectric conversion for each line.

  Therefore, the present invention has been made in view of such problems, and in the case of using an image sensor that sequentially captures image signals for each pixel (or for each line) such as a CMOS sensor, image blurring is caused. An object of the present invention is to provide an imaging apparatus capable of appropriately correcting.

  An imaging apparatus according to the present invention is an imaging apparatus that generates a multiplexed image by multiplexing and synthesizing captured images of two or more frames that are sequentially sequentially captured with respect to the same subject image. An image sensor that generates a captured image of one frame from an object image that is sequentially exposed and imaged by an imaging lens, and a feature point pixel of each captured image that corresponds to this feature part The image feature extraction means for identifying the image capturing means, the means for detecting the angular velocity of the imaging device body, and the feature portion pixels are exposed in each captured image. At least at least one of the captured images of two or more frames in order to correct the relative position shift of each captured image caused by the shake, and the motion detection means for detecting the generated motion. An image shifting means for shifting the photographic image of one frame, characterized in that it comprises a multiplexing image means for multiplexing the respective photographed image that shifted.

  The image feature extraction unit extracts, for example, a human face portion of the subject image as a feature portion. Further, the image feature extraction means detects, for example, a focal point position in the subject image, and extracts the focal point position as a feature portion. The image feature extraction unit extracts, for example, the position of the subject image to which the photographer's line of sight is directed as a feature portion. As a result, in the present invention, since the blur correction is performed so that the blur of the person's face, the in-focus position, or the position where the photographer's line of sight is directed is reduced, a more desirable image blur correction can be realized. it can.

  The image sensor is a CMOS sensor, for example.

  As described above, according to the present invention, the blur between the characteristic portions of the subject image is calculated as the blur between the captured images, so that it is possible to reduce the blur of the characteristic portions and realize a desirable image blur correction.

  Hereinafter, embodiments according to the present invention will be described with reference to the drawings. In the imaging apparatus according to the present embodiment, as will be described below, the first to third captured images that are sequentially captured with respect to the same subject image are multiplexed and synthesized, and the still image in which the multiplexed image is captured Is generated as Here, since camera shake normally occurs in the imaging apparatus, in order to correct the camera shake, each captured image is synthesized after being shifted by an amount corresponding to the camera shake. This will be specifically described below with reference to embodiments. In the embodiment described below, a case where the imaging device is a digital camera will be described, but the imaging device is not limited to a digital camera.

  FIG. 1 is a block diagram of a digital camera according to the first embodiment. The digital camera 10 according to the present embodiment includes an imaging device 20, and a subject image is formed on the imaging device 20 by an imaging lens 19. When a shutter button (not shown) is pressed and an image capturing operation is executed, the digital camera 10 continuously captures the same subject image, and the image sensor 20 captures a plurality of frames of captured images (in this embodiment, , First to third photographed images) are generated. The generated image signals are sequentially stored in the frame buffer 22 and then input to the image shift device 24.

  The digital camera 10 includes first and second angular velocity sensors 41 and 42 for detecting camera shake occurring in the camera body. Each of the first and second angular velocity sensors 41 and 42 is in a direction perpendicular to the z-axis serving as the optical axis and the surface composed of the z-axis serving as the optical axis and the x-axis extending in the horizontal direction at regular intervals (for example, 1 ms). Angular velocities Vx and Vy of the surface composed of the extending y-axis are detected. The horizontal direction is perpendicular to the optical axis direction and coincides with the horizontal direction in a normal use state, while the vertical direction is perpendicular to the optical axis direction and perpendicular to the horizontal direction.

  The first and second angular velocities vx and vy are input to the first and second A / D converters 51 and 52 as analog signals after amplifying signals relating to the angular velocities, respectively. The first and second A / D converters 51 and 52 perform A / D conversion on the first and second angular velocities vx and vy, and then store them as angular velocity data in the memory 62 via the angular velocity-motion vector converter 61. Once stored.

  In the angular velocity-motion vector conversion device 61, the first and second angular velocity data vx and vy stored in the memory 62 are read out, and the third and second photographed images with respect to the second and first photographed images caused by camera shake are generated. Are calculated respectively. The calculated misregistration data is temporarily stored in the memory 62 and then input to the image shift device 24.

  The exposure timing generator 71 generates an exposure start signal Pe and an exposure end signal Pf, and the image sensor 20 performs exposure in accordance with the timing of the exposure start signal Pe and the exposure end signal Pf. In the image feature extraction device 72, a feature point (feature portion) is extracted from the subject image, and one feature point pixel (feature portion pixel) on which the feature point is formed is specified. Further, in the image feature extraction device 72, the position information of the feature point pixel (the number of lines in which the feature point pixel is located and the number of pixels from the left side in the line) in each of the first to third photographed images. Calculated. In the angular velocity-motion vector conversion device 61, the timing at which the feature point pixel is exposed is calculated from this position information, and the adoption timing of the first and second angular velocities vx and vy is determined based on the calculated timing. Note that the image feature extraction device 72 in this embodiment includes a face recognition circuit, and extracts the center of a human face as a feature point of a subject image.

  In the image shift device 24, the second photographed image is shifted in the direction opposite to the displacement direction so as to correct the position deviation relative to the first photographed image based on the position deviation data. Input to the multiplexer 26. The third photographed image is shifted in the opposite direction to the misalignment direction so that the misalignment with respect to the second photographed image and the misalignment with respect to the first photographed image of the second photographed image are corrected. Is input to the image multiplexing device 26. Note that the first captured image is directly input to the image multiplexing device 26 without being shifted by the image shift device 24.

  In the image multiplexing device 26, the second photographed image is superimposed on the first photographed image, and the third photographed image is superimposed on the second photographed image. Here, as described above, since the second and third photographed images are overlaid after being shifted by the amount of displacement, the obtained multiplexed image is an image in which camera shake is corrected. The multiplexed image is stored as a still photographed image in the recording device 27 and displayed as a still photographed image on a monitor (not shown).

  FIG. 2 is a diagram for illustrating an image capture timing performed by the image sensor 20. FIG. 3 is a diagram showing the exposure period of each line and the relationship between the exposure start signal Pe and the exposure end signal Pf. In the present embodiment, the image sensor 20 is composed of a CMOS image sensor, and a subject image is formed by the imaging lens 19 on the imaging surface of the CMOS image sensor. The imaging surface of the imaging element 20 is perpendicular to the optical axis of the imaging lens 19, and as shown in FIG. 2, a horizontal line H formed by arranging pixels in the horizontal direction in the imaging region has n in the vertical direction. Line is provided.

  Each pixel of the CMOS image sensor includes a photodiode that senses incident light and generates a signal charge corresponding to the amount of incident light, a capacitor that accumulates the generated signal charge, and a plurality of MOSs that amplify and extract the signal charge from the capacitor. A transistor.

  As shown in FIG. 3, during the exposure of one frame, the exposure start signal Pe and the exposure end signal Pf are respectively equal to the number of lines of the CMOS image sensor (that is, the first to nth) at predetermined intervals. Exposure start signal Pe and first to nth exposure end signals Pf). As shown in FIG. 3, the first to nth exposure start signals Pe and the first to nth exposure end signals Pf are alternately input.

  In the CMOS image sensor according to this embodiment, photographing is performed using an electronic shutter, exposure is started in accordance with the input timing of the exposure start signal Pe, and exposure is ended in accordance with the input timing of the exposure end signal Pf. That is, when the first exposure start signal Pe is input, the line 1 of the horizontal line H is selected and selected during the period until the first exposure end signal Pf is input (unit exposure period E). In the horizontal line H, the pixels are further sequentially selected from left to right (see FIG. 2). Each exposure start signal Pe is input at the same timing in any frame. On the other hand, each exposure end signal Pf may be input at the same timing in any frame or may be input at a different timing. By inputting the exposure end signal Pf at a different timing for each frame, each unit exposure period E can be changed for each frame.

  Each pixel is exposed for the unit exposure period E, and by the exposure, electric charge is generated in each pixel by the photodiode and accumulated in the capacitor. The accumulated charge is read out as a pixel signal for each pixel and stored in the frame buffer 22. This exposure operation is similarly performed for each line from the line 2 to the last line (line n), whereby the exposure of the captured image for one frame is completed, and an image signal of the captured image for one frame is obtained. . Note that a period from the exposure start timing of the first line to the exposure end timing of the last line (line n), that is, an exposure period for one frame is referred to as a frame exposure period in this embodiment.

  FIG. 4 is a diagram specifically illustrating a method of extracting feature point pixels from the first to third captured images. As shown in FIG. 4, in the first to third photographed images, a face area Y with a face is extracted by a face recognition circuit. Specifically, the photographed image stored in the frame buffer 22 is read, and the skin color area in the photographed image is extracted. An average face image template is stored in the face recognition circuit, the average face image is compared with the skin color area, and a portion of the skin color area closest to the face image is extracted as the face area Y. The face recognition method is described in, for example, JP-A-8-63597. However, the face recognition method is not limited to the above method, and various other known methods can be used.

  In the present embodiment, the face area Y is extracted as a square area as shown in FIG. When the face area Y is extracted, the center pixel of the area Y (that is, the pixel located at the center of gravity of the area) is calculated, and the center pixel is detected as the feature point pixel C of each captured image.

  FIG. 5 is a graph showing the timing at which each operation in the photographing operation is performed over time. In FIG. 5, the vertical axis represents angular velocity, and the horizontal axis represents time. However, the angular velocity data is represented with respect to vx, and description of vy is omitted.

  Hereinafter, a method for calculating the positional deviation between the captured images from the angular velocity will be described with reference to FIG. As shown in FIG. 5, when the photographing operation is started, first, photographing of the first photographed image is started. That is, exposure is started in the image sensor 20, and exposure for capturing a captured image for one frame is performed in the first frame exposure period F1.

  When the first frame exposure period F1 ends, after the shooting interval I has elapsed, exposure for generating a second shot image is performed in the second frame exposure period F2. When the second frame exposure period F2 ends, after the shooting interval I has elapsed, exposure for generating a third shot image is performed in the third frame exposure period F3. Since the first to third photographed images are photographed under the same exposure conditions, the first to third frame exposure periods F1, F2, and F3 are the same as each other and the photographing interval I is also the same. is there.

  In the image feature extraction device 72, the feature point pixel C is detected as described above, and the feature point pixel C has timings A, A ′, A in each of the first to third photographed images as shown in FIG. The angular velocity-motion vector converter 61 calculates these timings A, A ′, A ″ from the position information of the feature point pixel C, and uses these timings A, A ′, A ″ to Specifically, the angular velocity data vx and vy detected in the period AA ′ are read from the memory 62. Each of the read angular velocity data vx and vy is: Time-integrated and calculated as the amount of rotation of the plane consisting of the z-axis and the x-axis and the plane consisting of the z-axis and the y-axis of the digital camera 10 in the period AA ′. flat Since the vertical plane with respect to the optical axis coincides with the imaging plane of the imaging device 20, the motion vector Vt1 is the fluctuation of the imaging plane that occurred during the period AA ′, that is, the feature point pixel C. This corresponds to a shake between each other, and is detected as a positional deviation between the first and second captured images.

  Similarly, the shake of the imaging surface (motion vector Vt2) in the period A′-A ″ is calculated from the angular velocity data vx and vy detected in the period A′-A ″, and the motion vector Vt2 is the second and third. It is detected as a positional deviation between captured images.

  As described above, the timings A and A ′ are timings when the center position of the face area Y of each captured image is exposed. In other words, the shake of the imaging surface (motion vector Vt1) that occurs during the period A-A ′ corresponds to the shake of the center position of the face between the first and second captured images. Here, the human face corresponds to the normal main image in the photographed image, and is the object for which image blur is most desired to be corrected. Therefore, when the blur at the center position of the face is detected as a positional shift (motion vector Vt1) between the first and second captured images, the detected positional shift is the positional shift of the target for which image blur correction is most desired. It becomes. Of course, the same can be said for the positional deviation detected during the period A′-A ″.

  FIG. 6 is a diagram schematically illustrating a method of multiplexing captured images in the image multiplexing device 26. 6 shows a configuration in which the second photographed image is multiplexed on the first photographed image, the third photographed image is similarly multiplexed. As shown in FIG. 6, in the image multiplexing device 26, the first captured image is first stored in the image multiplexing device 26, and after the first captured image is stored, the first captured image is stored on the first captured image. Two captured images are overlaid.

  Here, the second captured image is shifted by the image shift device 24 in a direction opposite to the direction of the motion vector Vt1 by the image shift device 24 relative to the first captured image by the scalar of the motion vector Vt1. 26, the image is multiplexed onto the first photographed image. The third photographed image is also an image shift device that is relative to the first photographed image by the scalar of the motion vector Vt1 + Vt2 in a direction opposite to the direction of the motion vector Vt1 + Vt2 obtained by adding the first motion vector and the second motion vector. 24 and is multiplexed on the second photographed image by the image multiplexing device 26.

  As described above, in the present embodiment, the blurring between the most characteristic portions (human faces) in the subject image is calculated as a positional deviation between the respective captured images, and each captured image is shifted based on the positional deviation. After that, the images are multiplexed. Therefore, in the obtained multiplexed image, the characteristic portion (human face) that is the object most desired to be corrected for blurring is most accurately corrected for blurring, so that desirable image blur correction can be realized.

  The second embodiment will be described with reference to FIGS. In the digital camera 10 according to the second embodiment, a focus detection device for detecting the focus of a subject has a plurality of distance measuring sensors, and the focus detection device has not only a function as focus detection but also image feature extraction. It also functions as a device. Hereinafter, the second embodiment will be described focusing on differences from the first embodiment.

  The focus detection device is composed of nine distance measuring sensors, and as shown in FIG. 7, detects which one of the focus detection points a to i is focused in the subject image OI, and a position where the focus is achieved. Is detected as the focal position.

  In the present embodiment, the focus is on the subject closest to the digital camera. Specifically, a phase difference image of the subject corresponding to the focus detection points a to i is formed from each distance measuring sensor, and the point having the largest phase difference among the focus detection points a to i in the subject image OI. That is, one point representing the subject closest to the camera is detected, and that one point is detected as the focal position. As shown in FIG. 7, three focus detection points are arranged in the row direction so as to be lowered to the left, and three rows are arranged in the column direction.

  FIG. 8 is a graph showing the timing at which each operation in the photographing operation of the second embodiment is performed over time. In the second embodiment, as described above, one of the focus detection points a to i is first detected as the in-focus position. Next, the position of the lens 19 is adjusted so that the subject at the in-focus position is in focus, and focus adjustment is performed. After the focus adjustment, the first to third captured images are captured. As in the first embodiment, the photographing operation is performed in the first to third frame exposure periods F1, F2, and F3, while the first to third photographed images are photographed and the first and second frame exposures are performed. A shooting interval I is provided between the periods F1 and F2 and the second and third frame exposure periods F2 and F3.

  In the image feature extraction device 72, first, a pixel corresponding to the focus detection point set at the in-focus position is detected as a feature point pixel, and position information of the feature point pixel is calculated. Next, the angular velocity-motion vector converter 61 determines the timing at which the feature point pixel is exposed from the position information of the feature point pixel. For example, when the focus detection point a is determined as the focus position, the timing at which the feature point pixel is exposed is the timing A, A ′, A ″ in each of the first to third captured images, as shown in FIG. Is calculated as

  In the angular velocity-motion vector conversion device 61, the angular velocities are further time-integrated in the periods AA ′ and A′-A ″ to calculate the motion vectors Vt1 and Vt2. The calculated motion vectors Vt1 and Vt2 are calculated. Is detected as a positional shift between the first and second photographed images (or the second and third photographed images), respectively, as in the first embodiment. Since it is the same as the form, the description is omitted.

  The in-focus position in the subject image corresponds to an important image portion in the photographed image, and is the image portion where image blur is most desired to be corrected. That is, in the present embodiment, by using the in-focus position as a feature point of an image, as in the first embodiment, it is possible to reduce blur of an important image portion and realize desirable image blur correction.

  Note that timings B, B ′, and B ″ shown in FIG. 8 are timings at which the pixels corresponding to the focus detection point b are exposed, and other timings C to I, C ′ to I ′, and C ″ to I ″. In the case where the focus detection points b to i are determined as in-focus positions, the positional deviation is calculated as described above.In the present embodiment, the focus is set on the subject closest to the camera. However, the focus may be adjusted to another position of the subject.

  Next, a third embodiment will be described with reference to FIGS. In the first embodiment, the image feature extraction device 72 is configured by a face recognition circuit. In the present embodiment, the image feature extraction device 72 is configured by a line-of-sight detection device.

  FIG. 9 is a diagram showing an outline of the visual line detection device. The line-of-sight detection device includes a light source (not shown), a light receiving element 80, a light receiving lens 81, and a light splitter 83. The light splitter 83 is provided inside the finder system so as to be positioned on the line of sight V of the photographer's eyeball, and transmits visible light and reflects infrared rays. When the photographer's eyeball 82 looks into the viewfinder system, the line of sight V passes through the light splitter 83 and is directed to the subject OB.

  When the photographer's eyeball 82 looks into the viewfinder system, the eyeball 82 is irradiated with infrared rays from a light source (not shown). In the eyeball 82, the infrared ray is reflected, and the reflected light is reflected by the light splitter 83 and then enters the light receiving element 80 through the light receiving lens 81.

  The light receiving element 80 detects the line of sight V using the reflected light. For example, the parallel light flux from the light source is projected onto the anterior segment of the eyeball 82 of the observer, whereby the direction and position of the pupil are obtained, and the line of sight V is calculated from the direction and position information. The method for detecting the line of sight is described in detail, for example, in Japanese Patent Laid-Open No. 6-86758. The line of sight detection is not limited to this method, and may be performed by any known method. In the line-of-sight detection device, when the line of sight V is detected, the position of the subject to which the line of sight V is directed is obtained, and the position is extracted as the feature point a.

  FIG. 10 is a graph showing the timing at which each operation in the shooting operation of the third embodiment is performed with time. In the third embodiment, when the above-mentioned line-of-sight detection is performed before shooting and the feature point a is extracted, the position information of the feature point pixel on which the feature point a is imaged is detected, and the position information is It is input to the angular velocity-motion vector converter 61.

  Thereafter, the first to third photographed images are photographed. As in the first embodiment, the photographing operation is performed in the first to third frame exposure periods F1, F2, and F3, while the first to third photographed images are photographed and the first and second frame exposures are performed. A shooting interval I is provided between the periods F1 and F2 and the second and third frame exposure periods F2 and F3.

  In the angular velocity-motion vector conversion device 61, the timings A, A ′, A ″ (see FIG. 10) at which the feature point pixels are exposed in each of the first to third captured images using the position information of the feature point pixels. Calculated.

  Also in the third embodiment, similarly to the first embodiment, the angular velocity data vx and vy detected in the periods AA ′ and A′-A ″ are time-integrated to calculate motion vectors Vt1 and Vt2. Then, the motion vectors Vt1 and Vt2 are detected as a positional deviation between the first and second captured images (or the second and third captured images) Since the following operations are the same as those in the first embodiment. The description is omitted.

  When a photographer shoots a subject image, the sight line V of the photographer peeking from the viewfinder system is directed to the subject portion that the photographer wants to photograph. That is, in the present embodiment, the subject portion to which the line of sight V is directed is used as a feature point of the image, thereby reducing blurring of the subject portion that the photographer wants to photograph and realizing desirable image blur correction.

  In the first embodiment, the feature point pixel is detected in each of the first to third captured images and may be different in each captured image. However, in the second and third embodiments, the feature of the image is different. Since point detection (that is, focus position detection, for example) is performed before shooting, and a feature point pixel is determined based on the feature point, the feature point pixel is the same in any of the first to third shot images. . However, also in the second and third embodiments, the line-of-sight detection and the focus position detection are performed every time each captured image is generated, and the feature point pixel may be different in each captured image. Similarly, also in the first embodiment, feature point detection may be performed before photographing, and the feature point pixels may be the same in any of the first to third photographed images.

1 is a block diagram of a digital camera according to a first embodiment. It is a figure for showing the capture timing of the image performed with an image sensor. It is a figure which shows the relationship between the exposure period of each line of an image pick-up element, an exposure start signal, and an exposure end signal. It is a figure for demonstrating the method in which the characteristic part is detected in the 1st thru | or 3rd picked-up image. It is a graph which shows the timing when each operation | movement in imaging | photography operation | movement is implemented with progress of time. It is the figure which showed typically the multiplexing method of the picked-up image in an image multiplexing apparatus. It is a figure for demonstrating the method by which the characteristic part is detected in 2nd Embodiment. It is a graph which shows the timing which each operation | movement is implemented with time passage in the imaging | photography operation | movement which concerns on 2nd Embodiment. It is a figure for showing the outline of a look detection device. It is a graph which shows the timing which each operation | movement is implemented with time passage in the imaging | photography operation | movement which concerns on 3rd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 20 Image sensor 24 Image shift apparatus 26 Image multiplexing apparatus 61 Angular velocity-motion vector conversion apparatus 71 Exposure timing generation apparatus 72 Image feature extraction apparatus

Claims (5)

An imaging apparatus that generates a multiplexed image by multiplexing and synthesizing captured images of two or more frames that are sequentially sequentially captured with respect to the same subject image.
An image sensor that sequentially exposes each pixel or line and generates a photographic image of one frame from the subject image formed by the imaging lens;
Image feature extraction means for extracting a feature portion of the subject image and specifying a feature portion pixel of each captured image corresponding to the feature portion;
Means for detecting an angular velocity of the imaging device body;
The feature detection pixel is exposed in each of the captured images, and during the timing, the angular velocity is time-integrated to detect blur generated between the captured images,
Image shift means for shifting a photographed image of at least one frame among the photographed images of two or more frames in order to correct a relative positional shift of each photographed image caused by the blur;
An imaging apparatus comprising: multiplexed image means for multiplexing the shifted captured images.   The imaging apparatus according to claim 1, wherein the image feature extraction unit extracts a human face portion of the subject image as the feature portion.   The imaging apparatus according to claim 1, wherein the image feature extraction unit detects a focal point position in the subject image and extracts the focal point position as the characteristic part.   The imaging apparatus according to claim 1, wherein the image feature extraction unit extracts a position of the subject image to which a photographer's line of sight is directed as a feature portion. The imaging apparatus according to claim 1, wherein the imaging element is a CMOS sensor.

Images