CN113596431A

CN113596431A - Image processing apparatus, image capturing apparatus, image processing method, and storage medium

Info

Publication number: CN113596431A
Application number: CN202110790632.2A
Authority: CN
Inventors: 福田浩一
Original assignee: Canon Inc
Current assignee: Canon Inc
Priority date: 2015-05-08
Filing date: 2016-04-21
Publication date: 2021-11-02
Anticipated expiration: 2036-04-21
Also published as: CN113596431B; WO2016181620A1

Abstract

The present invention provides an image processing apparatus, an image capturing apparatus, an image processing method, and a storage medium capable of generating a parallax image with improved quality. The image processing apparatus includes: an acquisition unit (125a) for acquiring a parallax image generated based on a signal of one of a plurality of photoelectric converters for receiving light beams passing through mutually different partial pupil areas in an imaging optical system, and acquiring a captured image generated by synthesizing a plurality of signals of the plurality of photoelectric converters; and an image processing unit (125b) for performing correction processing based on the captured image to reduce defects included in the parallax image.

Description

Image processing apparatus, image capturing apparatus, image processing method, and storage medium

(this application is a divisional application of an application having an application date of 2016, 4, 21, and an application number of 201680026847.4 entitled "image processing apparatus, image pickup apparatus, image processing method, and storage medium")

Technical Field

The present invention relates to an image processing apparatus capable of correcting a parallax image.

Background

Conventionally, there is known an image pickup apparatus capable of dividing an exit pupil of an image pickup lens into a plurality of pupil areas and simultaneously photographing a plurality of parallax images corresponding to the divided pupil areas. The photographed parallax image (viewpoint image) is equivalent to LF (light field) data which is information on the spatial distribution and angular distribution of light intensity.

Patent document 1 discloses an image pickup apparatus using a two-dimensional image pickup element including one microlens and a plurality of divided photoelectric converters for one pixel. The divided photoelectric converters receive light beams passing through different partial pupil areas of the exit pupil of the imaging lens via one microlens, thereby performing pupil division. A plurality of parallax images corresponding to the divided partial pupil areas may be generated based on the divided light reception signals of the respective photoelectric converters. Patent document 2 discloses an image pickup apparatus that adds up respective light reception signals of divided photoelectric converters and generates a captured image.

Documents of the prior art

List of cited documents

Patent document

Patent document 1: U.S. Pat. No. 4410804

Patent document 2: japanese patent laid-open No. 2001 + 083407

Disclosure of Invention

Problems to be solved by the invention

However, in a part of the parallax images (viewpoint images) obtained by the image pickup apparatuses disclosed in

patent documents

1 and 2, defects (defect signals) such as point defects and line defects, shadows and saturation signals caused by pupil division, and the like may be generated, and the parallax images may be degraded.

Accordingly, an object of the present invention is to provide an image processing apparatus, an image capturing apparatus, an image processing method, a program, and a storage medium, each of which can generate a parallax image with improved quality.

Means for solving the problems

As one aspect of the present invention, an image processing apparatus includes: an acquisition unit configured to acquire a parallax image generated based on a signal from one of a plurality of photoelectric converters for receiving light beams passing through mutually different partial pupil areas in an imaging optical system, and acquire a captured image generated by synthesizing a plurality of signals from the plurality of photoelectric converters; and an image processing unit configured to correct the parallax image based on the captured image.

As another aspect of the present invention, an image pickup apparatus includes: an image pickup element including a plurality of arrayed pixels provided with a plurality of photoelectric converters for receiving light beams passing through mutually different partial pupil regions in an imaging optical system; an acquisition unit configured to acquire a parallax image generated based on a signal from one of the plurality of photoelectric converters and acquire a captured image generated by synthesizing a plurality of signals from the plurality of photoelectric converters; and an image processing unit configured to correct the parallax image based on the captured image.

As another aspect of the present invention, an image processing method includes the steps of: acquiring a parallax image generated based on a signal from one of a plurality of photoelectric converters for receiving light beams passing through mutually different partial pupil areas in an imaging optical system, and acquiring a captured image generated by synthesizing a plurality of signals from the plurality of photoelectric converters; and correcting the parallax image based on the captured image.

As another aspect of the present invention, a program for causing a computer to execute a process including the steps of: acquiring a parallax image generated based on a signal from one of a plurality of photoelectric converters for receiving light beams passing through mutually different partial pupil areas in an imaging optical system, and acquiring a captured image generated by synthesizing a plurality of signals from the plurality of photoelectric converters; and correcting the parallax image based on the captured image.

As another aspect of the present invention, a storage medium stores the above-described program.

Other features and aspects of the present invention will become apparent from the following description of exemplary embodiments, which is to be read in connection with the accompanying drawings.

ADVANTAGEOUS EFFECTS OF INVENTION

The present invention can provide an image processing apparatus, an image capturing apparatus, an image processing method, a program, and a storage medium capable of generating a parallax image with improved quality.

Drawings

Fig. 1 is a block diagram of an image pickup apparatus in each embodiment.

Fig. 2 is a diagram showing a pixel array according to the first embodiment.

Fig. 3A is a diagram showing a pixel structure according to the first embodiment.

Fig. 3B is a diagram showing a pixel structure according to the first embodiment.

Fig. 4 is an explanatory diagram of an image pickup element and a pupil division function in each embodiment.

Fig. 5 is an explanatory diagram of an image pickup element and a pupil division function in each embodiment.

Fig. 6 is a diagram of the relationship between the defocus amount and the image shift amount in each embodiment.

Fig. 7A is a configuration diagram of a parallax image in each embodiment.

Fig. 7B is a configuration diagram of a parallax image in each embodiment.

Fig. 8 is a typical parallax image before correction processing is performed in the first embodiment.

Fig. 9 is a typical parallax image after correction processing in the first embodiment.

Fig. 10 is another typical parallax image before the correction processing is performed in the first embodiment.

Fig. 11 is another typical parallax image after the correction processing is performed in the first embodiment.

Fig. 12 is a diagram showing a pixel array in the second embodiment.

Fig. 13A is a diagram showing a pixel structure in the second embodiment.

Fig. 13B is a diagram showing a pixel structure in the second embodiment.

Fig. 14 is a schematic explanatory diagram of refocus processing in each embodiment.

Fig. 15A is an explanatory view of a light intensity distribution in the case where light is incident on the microlens formed on each pixel in the third embodiment.

Fig. 15B is an explanatory view of a light intensity distribution in the case where light is incident on the microlens formed on each pixel in the third embodiment.

Fig. 16 shows a light receiving rate distribution depending on the light incident angle in the third embodiment.

Fig. 17 is a schematic diagram of the flow of correction processing in the third embodiment.

Fig. 18 is a schematic diagram of the flow of the correction processing in the third embodiment.

Fig. 19A is an explanatory diagram of shading in the third embodiment.

Fig. 19B is an explanatory diagram of shading in the third embodiment.

Fig. 19C is an explanatory diagram of shading in the third embodiment.

Fig. 20A is an explanatory diagram of a projection signal of a captured image in the third embodiment.

Fig. 20B is an explanatory diagram of a projection signal of the first viewpoint image in the third embodiment.

Fig. 20C is an explanatory diagram of a shading function of a captured image in the third embodiment.

Fig. 21 is an example of a captured image (after demosaicing) in the third embodiment.

Fig. 22 is an example of the first viewpoint image before shading correction (after demosaicing) in the third embodiment.

Fig. 23 is an example of a shading-corrected first-viewpoint (first-corrected) image (after demosaicing) in the third embodiment.

Fig. 24 is an example of a first-viewpoint (first-corrected) image (after demosaicing and after shading correction) before defect correction in the third embodiment.

Fig. 25 is an example of a defect-corrected first-viewpoint (second-corrected) image (after demosaicing and after shading correction) in the third embodiment.

Fig. 26 is an example of a second-viewpoint (first-corrected) image (after demosaicing) before shading correction in the third embodiment.

Fig. 27 is an example of the shading-corrected second corrected viewpoint image (after demosaicing) in the third embodiment.

Description of reference numerals

125 image processing circuit (image processing apparatus)

125a acquisition unit

125b image processing unit

Detailed Description

Exemplary embodiments of the present invention will be described below with reference to the accompanying drawings.

First embodiment

A schematic structure of an image pickup apparatus according to a first embodiment of the present invention will now be described with reference to fig. 1. Fig. 1 is a block diagram of an image pickup apparatus 100 (camera) in the present embodiment. The image pickup apparatus 100 is a digital camera system including a camera body and an interchangeable lens (an imaging optical system or an image pickup optical system) removably mounted to the camera body. However, the present embodiment is not limited to this example, and is applicable to an image pickup apparatus including a camera body and a lens integrated with each other.

The first lens unit 101 is arranged on the frontmost side (object side) among a plurality of lens units constituting an imaging lens (imaging optical system), and is held on a lens barrel so as to reciprocate in the direction of the optical axis OA (optical axis direction). The stop/shutter 102 (aperture stop) adjusts its opening diameter to control the light amount when capturing an image, and also functions as a shutter to control the exposure time when capturing a still image. The second lens unit 103 reciprocates in the optical axis direction integrally with the stop/shutter 102, and has a zoom function of performing a magnification operation in conjunction with the reciprocation of the first lens unit 101. The third lens unit 105 is a focus lens unit that reciprocates in the optical axis direction to perform focusing (focusing operation). The optical low-pass filter 106 is an optical element for reducing false colors or moir e of a captured image.

The image pickup element 107 (image sensor) performs photoelectric conversion of an object image (optical image) formed by the imaging optical system, and includes, for example, a CMOS sensor or a CCD sensor and its peripheral circuits. As the image pickup element 107, for example, a two-dimensional single-plate color sensor provided with a primary color mosaic filter of a bayer array formed on a light receiving pixel having m pixels in the horizontal direction and n pixels in the vertical direction in an on-chip configuration is used.

The zoom actuator 111 rotationally moves (drives) an unillustrated cam barrel to move the first lens unit 101 and the second lens unit 103 in the optical axis direction, thereby performing a magnification varying operation. The stop/shutter actuator 112 controls the aperture diameter of the stop/shutter 102 to adjust the light amount (imaging light amount), and also controls the exposure time at the time of still image shooting. The focus actuator 114 moves the third lens unit 105 in the optical axis direction to perform focusing.

The electronic flash 115 is an illumination device to be used for illuminating an object. The electronic flash 115 may use a flash illumination device including a xenon tube or an illumination device including a continuously emitting LED (light emitting diode). The AF auxiliary light unit 116 projects an image of a mask having a predetermined opening pattern onto an object via a projection lens. This configuration can improve the focus detection capability for a dark object or an object with low contrast.

The CPU 121 is a control apparatus (control unit or controller) that manages various controls on the image capturing apparatus 100. The CPU 121 includes a processor, ROM, RAM, a/D converter, D/a converter, communication interface circuit, and the like. The CPU 121 reads out and executes a predetermined program stored in the ROM to drive various circuits of the image pickup apparatus 100, and performs a series of operations such as focus detection (AF), image pickup (shooting), image processing, or recording.

The electronic flash control circuit 122 controls the turning on and off of the electronic flash 115 in synchronization with the image capturing operation. The fill light driving circuit 123 controls turning on and off of the AF fill light unit 116 in synchronization with the focus detection operation. The image pickup element driving circuit 124 controls the image pickup operation of the image pickup element 107 and also performs a/D conversion of the acquired image signal to send it to the CPU 121.

An image processing circuit 125 (image processing apparatus) performs processing such as γ (gamma) conversion, color interpolation, or JPEG (joint photographic experts group) compression on the image data output from the image pickup element 107. In the present embodiment, the image processing circuit 125 includes an acquisition unit 125a and an image processing unit 125b (corrector). The acquisition unit 125a acquires a captured image and at least one parallax image (viewpoint image) from the image pickup element 107. The captured image is an image generated by synthesizing a plurality of signals (first and second signals) from a plurality of photoelectric converters (first and second sub-pixels) for receiving light beams passing through different partial pupil areas of the imaging optical system. The parallax image (viewpoint image) is an image generated based on signals from photoelectric converters (first sub-pixels or second sub-pixels) of the plurality of photoelectric converters. The image processing unit 125b performs correction processing (defect correction) based on the captured image so that defects contained in the parallax image are reduced.

A focus drive circuit 126 (focus driver) drives the focus actuator 114 based on the focus detection result to move the third lens unit 105 in the optical axis direction, thereby performing focusing. The stop/shutter drive circuit 128 drives the stop/shutter actuator 112 to control the aperture diameter of the stop/shutter 102. The zoom drive circuit 129 (zoom driver) drives the zoom actuator 111 in response to a zoom operation by the user.

The display device 131 (display unit) includes, for example, an LCD (liquid crystal display). The display device 131 displays information on the image capturing mode of the image capturing apparatus 100, a preview image before capturing, a confirmation image after capturing, a focus state display image at the time of focus detection, or the like. The operation member 132 (operation switch unit) includes a power switch, a release (image pickup trigger) switch, a zoom switch, an image pickup mode selection switch, and the like. The release switch is a two-stage switch having a half-pressed state (state in which SW1 is on) and a fully-pressed state (state in which SW2 is on). The recording medium 133 is, for example, a flash sensor removable with respect to the image capturing apparatus 100, and records a captured image (image data).

Now, a pixel array and a pixel structure of the image pickup element 107 according to the present embodiment will be described with reference to fig. 2, 3A, and 3B. Fig. 2 is a diagram illustrating a pixel array of the image pickup element 107. Fig. 3A and 3B are diagrams illustrating the pixel structure of the image pickup element 107, and fig. 3A illustrates a plan view (view viewed in the + z direction) of the pixel 200G of the image pickup element 107, and fig. 3B illustrates an a-a sectional view (view viewed in the a-z direction) in fig. 3A.

Fig. 2 shows a pixel array (array of image pickup pixels) on the image pickup element 107 (two-dimensional CMOS sensor) in a range of 4 columns × 4 rows. In the present embodiment, each image pickup pixel (

pixels

200R, 200G, and 200B) includes two

sub-pixels

201 and 202. Thus, FIG. 2 shows an array of subpixels of 8 columns by 4 rows.

As shown in fig. 2, a pixel group 200 of 2 columns × 2 rows includes

pixels

200R, 200G, and 200B in a bayer array. In the pixel group 200, the pixel 200R having the spectral sensitivity for R (red) is arranged at the upper left, the pixel 200G having the spectral sensitivity for G (green) is arranged at the upper right and lower left, and the pixel 200B having the spectral sensitivity for B (blue) is arranged at the lower right. Each of the

pixels

200R, 200G, and 200B (each image pickup pixel) includes sub-pixels 201 and 202 arranged in 2 columns × 1 rows. The sub-pixel 201 is a pixel for receiving a light beam passing through a first pupil area in the imaging optical system. The sub-pixel 202 is a pixel for receiving a light beam passing through a second pupil area in the imaging optical system.

As shown in fig. 2, the image pickup element 107 includes many image pickup pixels of 4 columns × 4 rows (subpixels of 8 columns × 4 rows) arranged on the surface thereof, and outputs image pickup signals (subpixel signals). In the image pickup element 107 of the present embodiment, the period P of the pixels (image pickup pixels) is 4 μm, and the number N of the pixels (image pickup pixels) is about 2075 ten thousand pixels, which is 5575 columns in the horizontal direction × 3725 rows in the vertical direction. In the image pickup element 107, the period P of the sub-pixels in the column direction_SUBIs 2 μm, and the number of sub-pixels N_SUBThe number of columns 11150 in the horizontal direction × rows 3725 in the vertical direction is about 4150 ten thousand pixels. Alternatively, the image pickup element 107 may have a pixel period P of 6 μm and a number N of pixels (image pickup pixels) of about 2400 ten thousand pixels in 6000 columns in the horizontal direction × 4000 rows in the vertical direction. Alternatively, in the image pickup element 107, the period P of the sub-pixels in the column direction_SUBMay be 3 μm, and the number of sub-pixels N_SUBIt may be 12000 columns in the horizontal direction by 4000 rows in the vertical direction, about 4800 ten thousand pixels.

As shown in FIG. 3B, the pixel 200G of this embodiment isThe light receiving surface side of the pixel is provided with a microlens 305 to condense incident light. The plurality of microlenses 305 are arranged in a two-dimensional manner, and each microlens 305 is arranged at a position distant from the light receiving surface by a predetermined distance in the z-axis direction (direction of the optical axis OA). In the pixel 200G, by dividing the pixel into N in the x direction_HTwo divided and divided into N in the y direction_VAnd (one division) to form a photoelectric converter 301 and a photoelectric converter 302 (photoelectric converter). The photoelectric converter 301 and the photoelectric converter 302 correspond to the sub-pixel 201 and the sub-pixel 202, respectively.

The

photoelectric converters

301 and 302 are each configured as a photodiode having a p-i-n structure including a p-type layer and an n-type layer and an intrinsic layer between the p-type layer and the n-type layer. The intrinsic layer may be omitted and a photodiode having a p-n junction may be applied, as desired. The pixel 200G (each pixel) is provided with a color filter 306 between the microlens 305 and each of the

photoelectric converters

301 and 302. The spectral transmittance of the color filter 306 may be changed for each sub-pixel, or alternatively, the color filter may be removed, as desired.

As shown in fig. 3A and 3B, light incident to the pixel 200G is condensed by the microlens 305, and is split by the color filter 306, and then the split light is received by the

photoelectric converters

301 and 302. In each of the

photoelectric converters

301 and 302, electron and hole pairs are generated according to the light receiving amount, and they are separated in the depletion layer, and then electrons having negative charges are accumulated in the n-type layer. On the other hand, holes are discharged to the outside of the image pickup element 107 via a p-type layer connected to a constant voltage source (not shown). The electrons accumulated in the n-type layers of the

photoelectric converters

301 and 302 are transferred to the electrostatic capacity portion (FD) via the transfer gate to be converted into a voltage signal.

Now, the pupil division function of the image pickup element 107 will be described with reference to fig. 4. Fig. 4 is an explanatory diagram of the pupil division function of the image pickup element 107, and shows pupil division of one pixel portion. Fig. 4 shows a-a cross-sectional view of the pixel structure shown in fig. 3A and an exit pupil of an imaging optical system, viewed from the + y direction. In fig. 4, the x-axis and the y-axis in the sectional view are inverted with respect to the x-axis and the y-axis of fig. 3A and 3B in order to correspond to the coordinate axes of the exit pupil plane.

In fig. 4, a partial pupil area 501 (first partial pupil area) of the sub-pixel 201 (first sub-pixel) has a substantially conjugate relationship with the light receiving surface of the photoelectric converter 301 whose center of gravity is shifted (decentered) in the-x direction via the microlens 305. Thus, the partial pupil area 501 represents a pupil area capable of receiving light by the sub-pixel 201. The center of gravity of the partial pupil area 501 of the sub-pixel 201 is shifted (decentered) in the + x direction on the pupil plane. A partial pupil region 502 (second partial pupil region) of the sub-pixel 202 (second sub-pixel) has a substantially conjugate relationship with the light receiving surface of the photoelectric converter 302 whose center of gravity is shifted (decentered) in the + x direction via the microlens 305. Thus, the partial pupil area 502 represents a pupil area that is capable of receiving light by the sub-pixel 202. The center of gravity of the partial pupil area 502 of the sub-pixel 202 is shifted (decentered) in the-x direction on the pupil plane. The pupil area 500 can receive light of the entire area of the pixel 200G when the photoelectric converters 301 and 302 (the sub-pixels 201 and 202) are fully combined.

Incident light is converged at a focal position by the microlens 305. However, the diameter of the condensed spot cannot become smaller than the diffraction limit Δ due to the influence of diffraction caused by the fluctuation of light, and it has a limited size. In the case where the light receiving faces of the

photoelectric converters

301 and 302 each have a length of about 1 to 2 μm, the converging spot of the microlens 305 is about 1 μm. Therefore, the

partial pupil regions

501 and 502 in fig. 4 having a conjugate relationship with the light receiving surfaces of the

photoelectric converters

301 and 302 via the microlens 305 are not clearly divided due to diffraction blur, and thus a light receiving rate distribution (pupil intensity distribution) is obtained.

Fig. 5 is an explanatory diagram of the image pickup element 107 and the pupil division function. Light beams that have passed through different

partial pupil areas

501 and 502 in the pupil area of the imaging optical system are incident on respective pixels of the image pickup element 107 on the image pickup plane 600 of the image pickup element 107 at angles different from each other, and are received by the sub-pixels 201 and 202 divided into 2 × 1 pieces. The present embodiment describes an example in which the pupil region is divided into two in the horizontal direction, but the present invention is not limited to the present embodiment, and pupil division may be performed in the vertical direction as needed.

In the present embodiment, the image pickup element 107 includes a plurality of sub-pixels that share one microlens and receive a plurality of light fluxes passing through different regions (a first partial pupil region and a second partial pupil region) in a pupil of an imaging optical system (image pickup lens). The image pickup element 107 includes a first sub-pixel (a plurality of sub-pixels 201) and a second sub-pixel (a plurality of sub-pixels 202) as a plurality of sub-pixels.

In this embodiment, signals of the sub-pixels 201 and 202 are added (synthesized) to each other for each pixel of the image pickup element 107 and read out so that a captured image having a resolution of the effective number of pixels N is generated. As described above, a captured image is generated by synthesizing the light reception signals of a plurality of sub-pixels (sub-pixels 201 and 202 in the present embodiment) for each pixel.

In the present embodiment, the light reception signals of the plurality of sub-pixels 201 are collected to generate the first parallax image. The first parallax image is subtracted from the captured image to generate a second parallax image. However, the present invention is not limited to this example, and the light reception signals of the plurality of sub-pixels 202 may be collected to generate the second parallax image. Thus, the parallax image is generated based on the light reception signals of the plurality of sub-pixels for the respective partial pupil regions different from each other.

In the present embodiment, the first parallax image, the second parallax image, and the captured image are each images employing a bayer array. Each of the first parallax image, the second parallax image, and the captured image may be demosaiced as necessary.

Now, the relationship between the defocus amount and the image shift amount of the first parallax image acquired from the sub-pixel 201 of the image pickup element 107 and the second parallax image acquired from the sub-pixel 202 will be described with reference to fig. 6. Fig. 6 shows a relationship between the defocus amount and the image shift amount. Fig. 6 illustrates the image pickup element 107 arranged on the image pickup plane 600, and as in fig. 4 and 5, the exit pupil of the imaging optical system is divided into two

partial pupil regions

501 and 502.

The defocus amount d is defined such that: the distance from the imaging position of the object to the image pickup surface 600 is | d |, the front focus state has a negative sign (d <0) at a position where the imaging position is located closer to the object side than the image pickup surface 600, and the back focus state has a positive sign (d >0) at a position where the imaging position is located on the opposite side of the object with respect to the image pickup surface 600. In an in-focus state in which the imaging position of the object is located on the imaging surface 600 (focus position), the defocus amount d is 0. Fig. 6 illustrates an object 601 in an in-focus state (d ═ 0) and an object 602 in a front focus state (d < 0). The front focus state (d <0) and the back focus state (d >0) are collectively referred to as a defocus state (| d | > 0).

In the front focus state (d <0), the light flux that has passed through the partial pupil area 501 (or the partial pupil area 502) among the light fluxes from the object 602 is once condensed. Then, the light flux is spread to a width Γ 1(Γ 2) centered at the center of gravity position G1(G2) of the light flux, and a blurred image is formed on the image pickup surface 600. The blurred image is received by the sub-pixel 201 (sub-pixel 202) constituting each pixel arranged in the image pickup element 107, and the first parallax image (second parallax image) is generated. Accordingly, the first parallax image (second parallax image) is recorded as a blurred object image in which the object 602 has a blur width Γ 1(Γ 2) at the barycentric position G1(G2) on the image pickup surface 600. The blur width Γ 1(Γ 2) of the object image increases substantially in proportion to an increase in the absolute value | d | of the defocus amount d. Likewise, the absolute value | p | of the image shift amount p of the object image between the first parallax image and the second parallax image, that is, equal to the difference (G1-G2) in the barycentric position of the light flux, increases substantially with the increase in the absolute value | d | of the defocus amount d. The same applies to the rear focus state (d >0), but the image shift direction of the object image between the first parallax image and the second parallax image is opposite to that in the front focus state.

As described above, according to the present embodiment, as the absolute value of the defocus amount of the first parallax image and the second parallax image or the defocus amount of the image signal obtained by adding the first parallax image and the second parallax image increases, the absolute value of the image shift amount between the first parallax image and the second parallax image increases.

Now, the correction processing of the parallax image in the present embodiment will be described. In this embodiment, the image pickup element 107 can output a captured image and at least one parallax image (at least one of the first parallax image and the second parallax image). The image processing circuit 125 (acquisition unit 125a) acquires the captured image and the parallax image output from the image pickup element 107. Then, the image processing circuit 125 (image processing unit 125b) corrects (corrects) the parallax image based on the captured image. The acquisition unit 125a may store the acquired captured image and the acquired at least one parallax image in a memory such as the recording medium 133 and the memory 134 to acquire the captured image and the parallax image stored in the memory, as necessary.

According to the circuit configuration or the driving manner of the image pickup element 107, due to the influence caused by the short circuit of the transfer gate, even if the captured image is normal, a defect signal (defect signal) may be generated in the parallax image (the first parallax image or the second parallax image), and defects such as a point defect and a line defect may be included in the parallax image. As necessary, defect information such as point defect information and line defect information inspected in a mass production process or the like may be stored in the memory in advance. In this case, the image processing circuit 125 (image processing unit 125b) performs correction processing of the parallax image by using the stored defect information. The image processing circuit 125 (checker) can check the parallax image in real time (i.e., during use of the image capturing apparatus 100 by the user) and determine defects such as point defects and line defects, as necessary.

Now, correction processing of a parallax image according to the present embodiment will be described with reference to fig. 7A and 7B. Fig. 7A is an array diagram of a parallax image (first parallax image) using a bayer array. Fig. 7B is an array diagram of a captured image using a bayer array. In fig. 7A and 7B, pixel values (pixel signals) of the first parallax image and the captured image at a position (j, I) of the jth pixel in the row direction and the ith pixel in the column direction are defined as a (j, I) and I (j, I), respectively.

If the first parallax image has a defect (line defect) on the jth line and the captured image is normal on the jth line, the jth line of the first parallax image needs to be corrected. In the present embodiment, the image processing circuit 125 (image processing unit 125b) corrects the first parallax image (pixel value at the position to be corrected in the first parallax image) based on the captured image. The second parallax image may be corrected similarly as necessary.

In the present embodiment, the correction value (correction signal) of the first parallax image at the position (j, i) where the defect occurs, that is, the position to be corrected (first position) is defined as Ac (j, i). The image processing unit 125b calculates a correction value Ac (j, i) according to the following expression (1), and corrects the first parallax image by using the calculated correction value Ac (j, i) as the pixel value a (j, i) of the first parallax image.

Expression 1

In expression (1), the parameters a0 and I0 are values for stabilizing the calculated values and suppressing an increase in noise in the case where the pixel value a of the first parallax image and the pixel value I of the captured image have low luminance (low luminance signal).

As described above, in the present embodiment, the image processing unit 125b performs correction processing of the parallax image based on the captured image, that is, replaces the pixel value a (j, i) of the parallax image at the position to be corrected with the correction value Ac (j, i). Specifically, the image processing unit 125b determines the correction value Ac (j, I) of the parallax image by using the pixel value I (j, I) of the captured image and the pixel values I (j2, I2) and a (j2, I2) of the captured image and the parallax image at the position (j2, I2) not (j, I) near the position to be corrected.

In the expression (1), the specific values of the parameters a0 and I0 may be set as appropriate. For example, if the pupil division number Np is 2, the parameters a0 and I0 may be set to a0 — I0/Np. The values of the parameters a0 and I0 may be changed according to image capturing conditions such as the position (j, I) to be corrected, ISO sensitivity, the aperture value of the imaging optical system, and the exit pupil distance. The values of the parameters a0 and I0 may be set based on the pixel value a of the first parallax image near (peripheral to) the position to be corrected and the pixel value I of the captured image.

Fig. 8 shows an example of the first parallax image (after demosaicing) in the in-focus state before the correction processing according to the present embodiment is performed. Fig. 9 shows an example of the first parallax image (after demosaicing) in the in-focus state after the correction processing according to the present embodiment is performed. Likewise, fig. 10 shows an example of the first parallax image (after demosaicing) in a defocused state before the correction processing is performed. Fig. 11 shows an example of the first parallax image in a defocused state after the correction processing is performed (after demosaicing). It is to be understood that the defect of the parallax image is corrected by the correction processing according to the present embodiment in each of the in-focus state and the out-of-focus state.

Now, refocus processing according to the present embodiment will be described with reference to fig. 14. Refocus processing is performed by the image processing circuit 125 (image processing unit 125b as a refocus unit) based on an instruction of the CPU 121. Fig. 14 is an explanatory diagram of refocus processing in a one-dimensional direction (column direction or horizontal direction) of a first signal (light reception signal of a first sub-pixel forming a first parallax image) and a second signal (light reception signal of a second sub-pixel forming a second parallax image) acquired by the image pickup element 107 according to the present embodiment. In fig. 14, symbol i denotes an integer, and schematic symbols Ai and Bi denote a first signal and a second signal, respectively, of the i-th pixel in the column direction of the image pickup element 107 arranged on the image pickup plane 600. The first signal Ai is a light reception signal output based on a light beam incident to the i-th pixel at a chief ray angle θ a (corresponding to the partial pupil area 501 in fig. 5). The second signal Bi is a light reception signal output as a light beam incident on the ith pixel at a chief ray angle θ b (corresponding to the partial pupil area 502 in fig. 5).

The first signal Ai and the second signal Bi each have incident angle information and light intensity distribution information. Therefore, the first signal Ai is translated (translated) to the virtual imaging plane 610 at an angle θ a and the second signal Bi is translated (translated) to the virtual imaging plane 610 at an angle θ b, which are then added to generate a refocusing signal on the virtual imaging plane 610. The parallel shift of the first signal Ai to the virtual imaging plane 610 by the angle θ a corresponds to a shift of +0.5 pixels in the column direction, and the parallel shift of the second signal Bi to the virtual imaging plane 610 by the angle θ b corresponds to a shift of-0.5 pixels in the column direction. Accordingly, in the case where the first signal Ai and the second signal Bi are relatively shifted by +1 pixel to add the first signal Ai and the corresponding second signal (Bi +1), a refocus signal on the virtual imaging plane 610 is generated. Likewise, in the case where the first signal Ai and the second signal Bi are shifted by an integral multiple of the pixel pitch (i.e., are integer-shifted) and these signals are added to each other, a shift addition signal (refocus signal) on each virtual imaging plane may be generated in accordance with the integer shift amount.

In the present embodiment, the influence of the defect included in at least one of the parallax images (at least one of the first parallax image and the second parallax image) is removed or reduced by the correction processing. Therefore, refocus processing can be performed based on the corrected parallax image. Therefore, refocus processing by using the respective signals (the first signal and the second signal) forming the parallax image can be performed with high accuracy.

Second embodiment

An image pickup apparatus according to a second embodiment of the present invention will now be described with reference to fig. 12 and fig. 13A and 13B. The present embodiment is different from the first embodiment in that: instead of generating the captured image based on the first parallax image and the second parallax image, the captured image is generated based on the first parallax image to the fourth parallax image which are a plurality of parallax images.

Fig. 12 illustrates a pixel array of the image pickup element 107 according to the present embodiment. Fig. 13A and 13B are diagrams illustrating the pixel structure of the image pickup element 107, and fig. 13A and 13B are a plan view (view from the + z direction) and a sectional view (view from the-z direction) along the line a-a of fig. 13A, respectively, of the pixel 200G of the image pickup element 107.

Fig. 12 shows a pixel array (array of image pickup pixels) of the image pickup element 107 (two-dimensional CMOS sensor) in a range of 4 columns × 4 rows. In the present embodiment, each image pickup pixel (

pixels

200R, 200G, and 200B) includes four

sub-pixels

201, 202, 203, and 204. Thus, fig. 12 shows an array of subpixels of 8 columns by 8 rows.

As shown in fig. 12, a pixel group 200 of 2 columns × 2 rows includes

pixels

200R, 200G, and 200B in a bayer array. In other words, in the pixel group 200, the pixel 200R having the spectral sensitivity for R (red) is arranged at the upper left, the pixel 200G having the spectral sensitivity for G (green) is arranged at the upper right and lower left, and the pixel 200B having the spectral sensitivity for B (blue) is arranged at the lower right. Each of the

pixels

200R, 200G, and 200B (each image pickup pixel) includes sub-pixels 201, 202, 203, and 204 arranged in 2 columns × 2 rows. The sub-pixel 201 is a pixel for receiving a light beam passing through a first pupil area of the imaging optical system. The sub-pixel 202 is a pixel for receiving a light beam passing through the second pupil area of the imaging optical system. The sub-pixel 203 is a pixel for receiving a light beam passing through the third pupil area of the imaging optical system. The sub-pixel 204 is a pixel for receiving a light beam passing through the fourth pupil area of the imaging optical system.

As shown in fig. 12, the image pickup element 107 includes many image pickup pixels (sub-pixels of 8 columns × 8 rows) of 4 columns × 4 rows arranged on the surface thereof, and outputs image pickup signals (sub-pixel signals). In the image pickup element 107 of the present embodiment, the period P of the pixels (image pickup pixels) is 4 μm, and the number N of the pixels (image pickup pixels) is about 2075 ten thousand pixels, which is 5575 columns in the horizontal direction × 3725 rows in the vertical direction. In the image pickup element 107, the period P of the sub-pixels in the column direction_SUBIs 2 μm, and the number of sub-pixels N_SUBThe number of columns 11150 in the horizontal direction × rows 7450 in the vertical direction is about 8300 ten thousand pixels. Alternatively, the image pickup element 107 may have a pixel (image pickup pixel) period P of 6 μm and a number N of pixels (image pickup pixels) of about 2400 ten thousand pixels which are 6000 columns in the horizontal direction × 4000 rows in the vertical direction. Optionally, the period P of the sub-pixels in the column direction_SUBMay be 3 μm, and the number of sub-pixels N_SUBIt may be 12000 columns in the horizontal direction by 4000 rows in the vertical direction, about 4800 ten thousand pixels.

As shown in fig. 13B, the pixel 200G of the present embodiment is provided with a microlens 305 on the light receiving surface side of the pixel to condense incident light. Each microlens 305 is disposed at a position distant from the light receiving surface by a predetermined distance in the z-axis direction (direction of the optical axis OA). In the pixel 200G, by dividing the pixel into N in the x direction_HTwo divided and divided into N in the y direction_VAnd (two-divided) to form

photoelectric converters

301, 302, 303, and 304 (photoelectric converters). The photoelectric converters 301 to 304 correspond to the sub-pixels 201 to 204, respectively.

In the present embodiment, the image pickup element 107 includes a plurality of sub-pixels that share one microlens and receive a plurality of light fluxes that pass through mutually different regions (first to fourth partial pupil regions) in the pupil of the imaging optical system (image pickup lens). The image pickup element 107 includes a first sub-pixel (a plurality of sub-pixels 201), a second sub-pixel (a plurality of sub-pixels 202), a third sub-pixel (a plurality of sub-pixels 203), and a fourth sub-pixel (a plurality of sub-pixels 204) as a plurality of sub-pixels.

In the present embodiment, signals of the sub-pixels 201, 202, 203, and 204 are added (synthesized) and read out for each pixel of the image pickup element 107 so that a captured image having a resolution of the effective number of pixels N is generated. As described above, the captured image is generated by combining the light reception signals of a plurality of sub-pixels (sub-pixels 201 to 204 in the present embodiment) for each pixel.

In the present embodiment, the light reception signals of the plurality of sub-pixels 201 are collected to generate the first parallax image. Likewise, light reception signals of the plurality of sub-pixels 202 are collected to generate a second parallax image, and light reception signals of the plurality of sub-pixels 203 are collected to generate a third parallax image. Further, in the present embodiment, the first parallax image, the second parallax image, and the third parallax image are subtracted from the captured image to generate the fourth parallax image. However, the present embodiment is not limited to this example, and light reception signals of the plurality of sub-pixels 204 may be collected to generate the fourth parallax image. As described above, the parallax image is generated based on the light reception signals of the plurality of sub-pixels for each of the partial pupil regions that are different from each other.

In the present embodiment, the captured image and the first to third parallax images (and the fourth parallax image) are all images employing a bayer array. Each of the captured image and the first to third parallax images (and the fourth parallax image) may be demosaiced as necessary. The correction processing (defect correction) of the parallax image according to the present embodiment is the same as that of the first embodiment, and therefore, the description thereof is omitted.

Third embodiment

Next, a third embodiment according to the present invention is explained. The present embodiment is different from the first embodiment in that: the image processing unit 125b performs light amount correction processing (shading correction) of the parallax image based on the captured image. In addition to the light amount correction processing according to the present embodiment, as in the first embodiment, correction processing of a parallax image may be performed based on a captured image to reduce defects contained in the parallax image.

The pupil area 500 shown in fig. 4 includes photoelectric converters 301 and 302 (divided into N) divided into 2 × 1 pieces via microlenses_x×N_yFirst to Nth photoelectric converters_LFPhotoelectric converter) has a substantially optically conjugate relationship with a light-receiving surface thereof. The pupil area 500 is a region each including sub-pixels 201 and 202 (first to nth sub-pixels)_LFSub-pixel) may receive a pupil area of light.

Fig. 15A and 15B are explanatory diagrams of light intensity distribution of microlenses formed when light is incident on each pixel. Fig. 15A shows a light intensity distribution of a cross section parallel to the optical axis of the microlens. Fig. 15B shows a light intensity distribution of a cross section perpendicular to the optical axis of the microlens. Incident light is converged to a focus position by a microlens. However, due to the influence of diffraction of the fluctuation of light, the diameter of the condensed spot cannot become smaller than the diffraction limit Δ and has a limited size. In the case where the light receiving surface of the photoelectric converter has a length of about 1 to 2 μm, the condensing spot of the microlens has a length of about 1 μm. Therefore, the

partial pupil regions

501 and 502 shown in fig. 4 having a conjugate relationship with the light receiving surface of the photoelectric converter via the microlens are not clearly pupil-divided due to diffraction blur, and have a light reception rate distribution (pupil intensity distribution) that depends on the light incident angle.

Fig. 16 is a diagram of a light reception rate distribution (pupil intensity distribution) depending on a light incident angle. The horizontal axis represents pupil coordinates, and the vertical axis represents light reception rate. A graph line L1 indicated by a solid line in fig. 16 represents a pupil intensity distribution along the x-axis in the partial pupil area 501 (first partial pupil area) of fig. 4. The light reception rate indicated by a graph line L1 rises sharply from the left end to a peak, then falls off gradually at a gentle rate, and then reaches the right end. A graph line L2 indicated by a broken line of fig. 16 represents a pupil intensity distribution along the x-axis of the partial pupil region 502 (second partial pupil region). In contrast to the graph line L1, the light reception rate indicated by the graph line L2 sharply increases from the right end, reaches its peak, gradually decreases at a gentle rate of change, and then reaches the left end. It is to be understood that, as shown in fig. 16, gentle pupil division is performed.

As shown in fig. 5, photoelectric converters 301 and 302 (first to nth photoelectric converters)_LFPhotoelectric converter) and sub-pixels 201 and 202 (first to nth sub-pixels)_LFSub-pixels) correspond. In each pixel of the image pickup element, subpixels 201 and 202 (divided into N) divided into 2 × 1 pixels_x×N_yFirst to Nth photoelectric converters_LFPhotoelectric converter) receives the light passing through the partial pupil areas 501 and 502 (first to nth sub-pixels)_LFSub-pixels) of different partial pupil areas. LF data (input image)) representing the spatial distribution and angular distribution of light intensity is obtained based on the signal received by each sub-pixel.

For each pixel, sub-pixels 201 and 202 (N) from 2 × 1 division can be passed based on LF data (input image)_x×N_yThe divided first to Nth photoelectric converters_LFPhotoelectric converters) are combined with each other to generate a captured image having a resolution of the number of pixels N.

Based on the LF data (input image), for each pixel, a sub-pixel from the 2 × 1 division is selectedPixels 201 and 202 (N)_x×N_yThe divided first to Nth photoelectric converters_LFA photoelectric converter) of a specific sub-pixel. Thereby, the partial pupil areas 501 and 502 (first to nth sub-pixels) can be generated_LFSub-pixels) corresponding to a specific partial pupil area. For example, a first viewpoint image (first parallax image) having a resolution of the number of pixels N corresponding to the partial pupil area 501 in the imaging optical system can be generated by selecting a signal from the sub-pixel 201. The same applies to the other sub-pixels.

As discussed, the image pickup element according to the present embodiment includes a plurality of array pixels having a plurality of photoelectric converters configured to receive light beams that have passed through different partial pupil areas of the imaging optical system, and can acquire LF data (input data). The present embodiment deals with the first viewpoint image and the second viewpoint images (first viewpoint image to nth viewpoint image)_LFViewpoint image) performs image processing (correction processing) such as stain correction and shading correction, and generates an output image.

Now, description will be given of a method for imaging a first viewpoint image and a second viewpoint image (first viewpoint image to nth viewpoint image) by using a captured image and LF data (input image) acquired by the image pickup element 107 with reference to fig. 17 and 18_LFViewpoint image) to perform correction processing to generate an output image. Fig. 17 and 18 are schematic diagrams of the flow of the correction processing according to the present embodiment. Fig. 17 and 18 are mainly executed by the image processing circuit 125 (the acquisition unit 125a and the image processing unit 125b) based on a command of the CPU 121.

First, at a stage before step S1 of fig. 17 (or at step S0 not shown), the image processing circuit 125 (acquisition unit 125a) generates (acquires) a captured image and at least one viewpoint image based on LF data (input data) acquired by the image pickup element 107. The captured image is an image generated from pupil areas obtained by combining different partial pupil areas in the imaging optical system. The viewpoint images are images generated for respective different partial pupil areas in the imaging optical system.

In step S0, first, the image processing circuit 125 inputs LF data (input image) acquired by the image pickup element 107. Alternatively, the image processing circuit 125 may use the LF data (input image) captured in advance by the image pickup element 107 and stored in a recording medium.

Next, in step S0, the image processing circuit 125 generates a captured image corresponding to a pupil area in which different partial pupil areas (first partial pupil area and second partial pupil area) in the imaging optical system are synthesized. The LF data (input image) is referred to as LF. In addition, the i-th pixel signal in the column direction in each pixel signal of the LF is set_sEach (1 is less than or equal to i)_s≤N_x) And j in the row direction_SIs (1 is less than or equal to j)_SNy) is called a k-th sub-pixel signal, where k is N_x(j_S-1)+i_S(1≤k≤N_LF). The image processing circuit 125 generates a synthetic image as the ith image in the column direction as expressed by the following expression (2)_SJ in the row direction_SThe photographed image I (j, I) of the person.

Expression 2

In order to maintain good S/N of the captured image (j, i), the present embodiment synthesizes the respective sub-pixel signals of expression (2) with each other in the electrostatic capacity section (FD) within the image pickup element before performing analog-to-digital conversion (a/D conversion) on the respective sub-pixel signals. As necessary, the present embodiment can synthesize the respective sub-pixel signals of expression (2) with each other when converting the electric charges accumulated in the electrostatic capacity section (FD) inside the image pickup element into voltage signals, before a/D conversion is performed on the respective sub-pixel signals. The present embodiment can synthesize the respective sub-pixel signals of expression (2) with each other after a/D conversion of the respective sub-pixel signals, as necessary.

In the present embodiment, each pixel is divided into two in the X direction (for example, N)_x＝2，N _y1 and N_LF2). Based on an input image (LF data) corresponding to the exemplary pixel array in fig. 2, sub-pixels 201 and 202(N is performed) obtained by binary division in the X direction are provided for each pixel_x×N_yDivided first to Nth sub-pixels_LFSub-pixels) are synthesized. This can be generated as a pixel having the number N (horizontal pixel number N)_HX number of vertical pixels N_V) A captured image of RGB signals of the bayer array of resolution (c). Since the correction processing of the viewpoint image according to the present embodiment uses the captured image as a reference image of the correction reference, shading (light amount) correction processing, point flaw correction processing, and the like are performed with respect to the captured image I (j, I). Other processing may be performed if desired.

Next, in step S0, the image processing circuit 125 generates a k-th viewpoint image I as the I-th in the column direction and the j-th in the row direction corresponding to the k-th partial pupil area of the imaging optical system as shown in the following expression (3)_k(j,i)。

Expression 3

In the present embodiment, each pixel is divided into two in the X direction (for example, N)_x＝2，N _y1 and N_LF2) and k 1. The present embodiment selects a signal of the sub-pixel 201, which is obtained by two-division in the x direction, for each pixel based on LF data (input image) corresponding to the pixel array shown in fig. 2. Then, the present embodiment is based on the partial pupil regions 501 and 502 (first to nth partial pupil regions)_LFPartial pupil area) as a pixel having the number of pixels N (horizontal pixel number N)_HX number of vertical pixels N_V) First viewpoint image I of RGB signals of bayer array of resolution₁(j, i). As necessary, k may be selected to be 2, and the second viewpoint image I corresponding to the partial pupil region 502 in the imaging optical system may be generated₂(j,i)。

As described above, the image processing unit 125b generates captured images corresponding to pupil areas resulting from synthesizing different partial pupil areas based on input images acquired by the image pickup element having a plurality of pixels configured to receive light beams having passed through the plurality of photoelectric converters of the different partial pupil areas in the imaging optical system. In addition, the image processing unit 125b generates at least one viewpoint image for each different partial pupil region.

The present embodiment generates a captured image I (j, I) as an RGB signal employing a bayer array and a first-viewpoint image I as an RGB signal employing a bayer array based on LF data (input image) acquired by the image pickup element 107₁(j, i) and storing them in a recording medium. In addition, the present embodiment is based on the captured image I (j, I) and the first viewpoint image I₁(j, I) generating a second view image I₂(j, i). With respect to the captured image I (j, I) in the present embodiment, this configuration can provide the same image processing as that with respect to a captured image acquired by a conventional image pickup element in which the photoelectric converter of each pixel is not divided. If necessary, the first viewpoint image I may be generated so that the processing for each viewpoint image is the same₁(j, I) and a second view image I₂(j, i) and storing it in a recording medium.

Next, in step S1 of fig. 17, the image processing unit 125b performs processing for the first viewpoint image I based on the captured image I (j, I)₁(kth view image I_k) Shading correction processing (light amount correction processing) for each of RGB.

Now, description will be given of the first and second viewpoint images (first to nth viewpoint images) with reference to fig. 19A to 19C_LFViewpoint image) of the image. Fig. 19A to 19C are explanatory diagrams of shading, and show the relationship between a partial pupil area 501 where the photoelectric converter 301 receives light, a partial pupil area 502 where the photoelectric converter 302 receives light, and the exit pupil 400 in the imaging optical system at the peripheral image height of the image pickup element 107. Using the same reference numerals to designateCorresponding elements in fig. 4. Photoelectric converters 301 and 302 (first to nth photoelectric converters)_LFPhotoelectric converter) and sub-pixels 201 and 202 (first to nth sub-pixels)_FLSub-pixels) correspond.

Fig. 19A illustrates a case where the exit pupil distance Dl in the imaging optical system is equal to the set pupil distance Ds in the image pickup element 107. In this case, the exit pupil 400 in the imaging optical system is divided approximately equally by the

partial pupil regions

501 and 502. In contrast, as shown in fig. 19B, in the case where the exit pupil distance Dl in the imaging optical system is shorter than the set pupil distance Ds in the image pickup element 107, pupil shift occurs between the exit pupil 400 and the entrance pupil of the image pickup element 107 at the peripheral image height of the image pickup element 107. As a result, the exit pupil 400 is not divided equally. Likewise, as shown in fig. 19C, in the case where the exit pupil distance Dl in the imaging optical system is longer than the set pupil distance Ds in the image pickup element 107, pupil shift occurs between the exit pupil 400 and the entrance pupil of the image pickup element 107 at the peripheral image height of the image pickup element 107. As a result, the exit pupil 400 is not divided equally. With unequal pupil division at the peripheral image height, the first viewpoint image and the second viewpoint image have unequal intensities and produce shadows each of which one of the first viewpoint image and the second viewpoint image is high in intensity and the other is low in intensity for RGB (color).

In order to generate each viewpoint image with good image quality, the image processing unit 125b according to the present embodiment performs processing for the first viewpoint image I (j, I) by using the captured image I (j, I) as a reference or reference image in step S1 of fig. 17₁(kth view image I_k) Shading correction (light amount correction) of each of RGB.

In step S1 of fig. 17, the image processing circuit 125 first detects the captured image I (j, I) and the first viewpoint image I₁Any of (j, i) is an unsaturated and defect-free (non-defective) effective pixel V₁(j, i). Photographed image I (j, I) and first-viewpoint image I₁An effective pixel in which any of (j, i) is unsaturated and is free of defects (non-defective) satisfies V₁(j, i) ═ 1. Photographed image I (j, I) and first-viewpoint image I₁An invalid pixel in which any of (j, i) is saturated or defective satisfies V₁(j, i) ═ 0. Similarly, for the k-th view image I_kIn the shading (light amount) correction of (1), the image I (j, I) and the k-th viewpoint image I are photographed_k(j, i) both of which are unsaturated and have no defect satisfy V_k(j,i)＝1。

Assume integer j₂(1≤j₂≤N_V/2) and i₂(1≤i₂≤N_H/2). It is assumed that the photographic image I employing the bayer array in fig. 2 includes photographic images RI, Gr, GbI, and BI for R, Gr, Gb, and B. The captured image of R is represented as RI (2 j)₂-1,2i₂-1)＝I(2j₂-1,2i₂-1), and the captured image of Gr is represented as GrI (2 j)₂-1,2i₂)＝I(2j₂-1,2i₂). The captured image of Gb is represented as GbI (2 j)₂,2i₂-1)＝I(2j₂,2j₂-1), and representing the captured image of B as BI (2 j)₂,2i₂)＝I(2j₂,2i₂). Likewise, assume that the k-th captured image I shown in fig. 2_kIncluding photographed images RI for R, Gr, Gb, and B_k、GrI_k、GbI_kAnd BI_k. Representing the captured image of R as RI_k(2j₂-1,2i₂-1)＝I_k(2j₂-1,2i₂-1), and representing the captured image of Gr as GrI_k(2j₂-1,2i₂)＝I_k(2j₂-1,2i₂). Representing the captured image of Gb as GbI_k(2j₂,2i₂-1)＝I_k(2j₂,2j₂-1), and representing the captured image of B as BI_k(2j₂,2i₂)＝I_k(2j₂,2i₂)。

In step S1, the image processing unit 125b then performs processing for the captured image RI (2 j)₂-1,2i₂-1)、GrI(2j₂-1,2i₂)、GbI(2j₂,2i₂-1) and BI (2 j)₂,2i₂) The projection processing of (1).More specifically, it is directed to the captured image RI (2 j)₂-1,2i₂-1)、GrI(2j₂-1,2i₂)、GbI(2j₂,2i₂-1) and BI (2 j)₂,2i₂) With expressions (4A) to (4D), projection processing is performed in a direction (y direction) perpendicular to the pupil-dividing direction (x direction). Thereby, a projection signal RP (2 i) of the captured image is generated₂-1)、GrP(2i₂)、GbP(2i₂-1) and BP (2 i)₂). The saturation signal value or the defect signal value does not contain information for detecting the shading of each of RGB for the captured image with correction. Thus, the captured image and the effective pixel V are calculated_kAnd projection processing (numerator of upper stage in expressions (4A) to (4D)) is performed by excluding the saturation signal value and the defect signal value, and normalization is performed with the effective pixel number for projection processing (denominator of upper stage in expressions (4A) to (4D)). In the case where the number of effective pixels used for the projection processing is 0, the projection signal of the photographing signal is set to 0 through the lower stage in expression (4A) to the lower stage in expression (4D). In the case where the projection signal of the captured image becomes a negative signal due to the influence of noise, the projection signal of the captured image is also set to 0. Also, for the k view image RI_k(2j₂-1,2i₂-1)、GrI_k(2j₂-1,2i₂)、GbI_k(2j₂,2i₂-1) and BI_k(2j₂,2i₂) With expressions (4E) to (4H), projection processing is performed in a direction (y direction) perpendicular to the pupil-dividing direction (x direction). This configuration generates a projection signal RP of the k-th viewpoint image_k(2i₂-1)、GrP_k(2i₂)、GbP_k(2i₂-1) and BP_k(2i₂)。

Expression 4A

Expression 4B

Expression 4C

Expression 4D

Expression 4E

Expression 4F

Expression 4G

Expression 4H

After the projection processing of expressions (4A) to (4H), low-pass filter processing is performed. Projection signal RP (2 i) for captured image₂-1)、GrP(2i₂)、GbP(2i₂-1) and BP (2 i)₂) And a projection signal RP of the k-th viewpoint image_k(2i₂-1)、GrP_k(2i₂)、GbP_k(2i₂-1) and BP_k(2i₂) And performing low-pass filtering processing. Thereby, the projection signal in the captured image is smoothed. Alternatively, the low-pass filtering process may be omitted.

Fig. 20A to 20C are explanatory diagrams of the projection signal of the captured image, the projection signal of the first viewpoint image, and the shading function. Fig. 20A shows examples of projection signals rp (r), grp (g), gbp (g), and bp (b) of the captured image. Fig. 20B shows a projection signal RP of the first viewpoint image₁(R)、GrP₁(G)、GbP₁(G) And BP₁(B) Examples of (2). Each projection signal has a plurality of fluctuations depending on the object. For the first view point image I₁(kth view image I_k) To perform shading correction with high accuracy, it is necessary to correct the first-viewpoint image I due to pupil shift₁(kth view image I_k) The respective shadow components of RGB (red, green, blue) are separated from the signal components of RGB for the subject.

In step 1, the image processing unit 125b calculates the k-th viewpoint image I with respect to the captured image as a reference by expressions (5A) to (5D)_kRespective shadow signals RS of RGB_k(2i₂-1)、GrS_k(2i₂)、GbS_k(2i₂-1) and BS_k(2i₂)。

Expression 5A

Expression 5B

Expression 5C

Expression 5D

In calculating the shading component, it is necessary to make the light-receiving amount of the pixel larger than that of the sub-pixel, and make the light-receiving amount of the sub-pixel larger than 0. Thus, when the conditional expression RP (2 i) is satisfied₂-1)>RP_k(2i₂-1)>In the case of 0, the R projection signal RP of the k-th viewpoint image is acquired by expression (5A)_k(2i₂-1) and the R projection signal RP (2 i) of the captured image₂-1) of the ratio between. Then, the result is compared with the pupil division number N_LFThe multiplication is performed for normalization, and a k-th viewpoint image I is generated_kR shadow signal RS_k(2i₂-1). Thereby, the R signal component of the subject can be canceled, and the k-th viewpoint image I can be separated_kR shadow component of (a). When the conditional expression RP (2 i) is not satisfied₂-1)>RP_k(2i₂-1)>In the case of 0, the k-th view image I_kR shadow signal RS_k(2i₂-1) is set to 0.

Likewise, when conditional expression GrP (2 i) is satisfied₂)>GrP_k(2i₂)>In the case of 0, the Gr projection signal GrP of the k-th viewpoint image is acquired by expression (5B)_k(2i₂) And Gr projection signal GrP (2 i) of the captured image₂) The ratio of (a) to (b). Then, the result is compared with the pupil division number N_LFThe multiplication is performed for normalization, and a k-th viewpoint image I is generated_kGr shadow signal GrS_k(2i₂). Thereby, the Gr signal component of the subject can be canceled, and the k-th viewpoint image I can be separated_kThe Gr shade component of (a). When conditional expression GrP (2 i) is not satisfied₂)>GrP_k(2i₂)>In the case of 0, the k-th view image I_kGr shadow signal GrS_k(2i₂) Is set to 0.

Likewise, when conditional expression GbP (2 i) is satisfied₂-1)>GbP_k(2i₂-1)>In the case of 0, the Gb projection signal GbP of the k-th viewpoint image is acquired by expression (5C)_k(2i₂-1) and a Gb projection signal GbP (2 i) of the captured image₂-1) of the ratio between. Then, the result is compared with the pupil division number N_LFThe multiplication is performed for normalization, and a k-th viewpoint image I is generated_kGb shadow signal GbS_k(2i₂-1). This makes it possible to cancel the Gb signal component of the subject and separate the k-th viewpoint image I_kGb shading component of (a). When conditional expression GbP (2 i) is not satisfied₂-1)>GbP_k(2i₂-1)>In the case of 0, the k-th view image I_kGb shadow signal GbS_k(2i₂-1) is set to 0.

Likewise, when conditional expression BP (2 i) is satisfied₂)>BP_k(2i₂)>In the case of 0, the B projection signal BP of the k-th viewpoint image is acquired by expression (5D)_k(2i₂) And B projection signal BP (2 i) of the captured image₂) The ratio of (a) to (b). Then, the result is compared with the pupil division number N_LFThe multiplication is performed for normalization, and a k-th viewpoint image I is generated_kB shadow signal BS_k(2i₂). Thereby, the B signal component of the subject can be canceled, and the k-th viewpoint image I can be separated_kB shadow component of (a). When conditional expression BP (2 i) is not satisfied₂)>BP_k(2i₂)>In the case of 0, the k-th view image I_kB shadow signal BS_k(2i₂) Is set to 0.

For high-precision shadow correction, the method satisfies the requirement of RS_k(2i₂-1)>0、GrS_k(2i₂)>0、GbS_k(2i₂-1)>0 and BS_k(2i₂)>In the case where the number of effective shadow signals of 0 is equal to or higher than the predetermined number, the shadow correction may be provided.

Next, in step S1, the image processing unit 125b performs the calculation processing represented by expressions (6A) to (6D). Will be for the k viewImage I_kRespective shadow function RSF of RGB_k(2i₂-1、GrSF_k(2i₂)、GbSF_k(2i₂-1) and BSF_k(2i₂) N set as smoothing for positional variation in pupil division direction (x direction)_SFA polynomial function of degree. RS that is generated by expressions (5A) to (5D) and satisfies RS_k(2i₂-1)>0、GrS_k(2i₂)>0、GbS_k(2i₂-1)>0 and BS_k(2i₂)>The valid shadow signal of 0 is set to the data point. The coefficient RSC in expressions (6A) to (6D) is calculated from these data points and parameter fitting by the least square method_k(μ)、GrSC_k(μ)、GbSC_k(. mu.) and BSCk (. mu.).

Expression 6A

Expression 6B

Expression 6C

Expression 6D

As described above, the k-th viewpoint image I for the captured image as a reference is generated_kRespective shadow function RSF of RGB_k(2i₂-1、GrSF_k(2i₂)、GbSF_k(2i₂-1)And BSF_k(2i₂)。

Fig. 20C shows the first viewpoint image I for the captured image as a reference₁Respective shadow function RSF of RGB₁(R)、GrSF₁(G)、GbSF₁(G) And BSF₁(B) In that respect The projection signal of the first viewpoint image in fig. 20B and the projection signal of the captured image in fig. 20A have subject-dependent fluctuations. On the other hand, fluctuations depending on the object (signal values of the object for each of RGB) can be cancelled out by obtaining the ratio between the projection signal of the first viewpoint image and the projection signal of the captured image, and the smoothed first viewpoint image I can be separated and generated for each of RGB₁The shading function of (a). Although the present embodiment uses a polynomial function as the shading function, the present invention is not limited to the present embodiment, and a more general function may be used according to the shading shape as needed.

Next, in step S1 of fig. 17, the image processing unit 125b targets the k-th viewpoint image I_k(j, i) with expressions (7A) to (7D), shading (light amount) correction processing is performed using the shading functions of RGB, respectively. Thereby, a shading-corrected k-th viewpoint (first corrected) image M is generated₁I_k(j, i). The k-th viewpoint (first corrected) image M employing the bayer array is represented as follows for each of R, Gr, Gb, and B₁I_k. In other words, the k-th viewpoint (first correction) image of R is set to RM₁I_k(2j₂-1,2i₂-1)＝M₁I_k(2j₂-1,2i₂-1), and the kth viewpoint (first correction) image of Gr is set to GrM₁I_k(2j₂-1,2i₂)＝M₁I_k(2j₂-1,2i₂). Set the kth viewpoint (first correction) image of Gb to GbM₁I_k(2j₂,2i₂-1)＝M₁I_k(2j₂,2i₂-1), and the kth viewpoint (first correction) image of B is set as BM₁I_k(2j₂,2i₂)＝M₁I_k(2j₂,2i₂). According to the needTo do so, the shading-corrected kth viewpoint (first corrected) image M may be₁I_k(j, i) is set as an output image.

Expression 7A

Expression 7B

Expression 7C

Expression 7D

Now, description will be made with reference to FIGS. 21 to 23 regarding the first viewpoint image I shown in step S1 of FIG. 17₁And (j, i) shading correction processing (light amount correction processing) for each of the RGB. Fig. 21 shows an example of a captured image I (after demosaicing) according to the present embodiment. The exemplary captured image has good image quality. FIG. 22 shows the first-viewpoint image I before shading correction in the present embodiment₁(after demosaicing) example. RGB each generate a shadow due to pupil shift between an exit pupil of an imaging optical system and an entrance pupil of an image pickup element, and an image I at a first viewpoint₁To the right of (j, i), the brightness decreases and the RGB ratio is modulated. Fig. 23 shows a first-viewpoint (first-corrected) image M after shading correction in the present embodiment₁I₁(after demosaicing) example. Brightness due to shading correction based on each of RGB of photographed imageThe modulation of the RGB ratio is reduced and corrected, and a shading-corrected first-viewpoint (first-corrected) image M having good image quality is generated as in the captured image₁I₁(j, i) (after demosaicing).

The present embodiment generates a captured image corresponding to a pupil area obtained by combining different partial pupil areas based on an input image acquired by an image pickup element including a plurality of pixels having a plurality of photoelectric converters configured to receive light beams having passed through the different partial pupil areas of an imaging optical system. Then, the present embodiment generates a plurality of viewpoint images for respective different partial pupil regions, performs image processing for correcting the viewpoint images based on the captured images, and generates an output image. The present embodiment performs image processing for correcting the light amounts (shades) for each color or RGB based on a captured image. The present embodiment performs image processing for correcting the light amount of the viewpoint image based on the projection signal of the captured image and the projection signal of the viewpoint image. The configuration of the present embodiment can provide a viewpoint image with good quality.

Next, in step S2 of fig. 17, the image processing unit 125b corrects the k-th viewpoint (first correction) image M after shading correction based on the captured image I₁I_kThe defect of (2). The present embodiment shows an example where k is 1, but the present invention is not limited to the present embodiment.

In the present embodiment, although the captured image I is normal, the k-th viewpoint image I is caused due to a short circuit of a transfer gate or the like caused by the circuit configuration and the driving manner of the image pickup element_k(first view image I₁) Only a portion of which generates a defect signal or a point or line defect. As necessary, the present embodiment may store point defect information and line defect information detected in a mass production process or the like in the image processing circuit 125 or the like in advance, and may perform processing for the k-th viewpoint image I based on the recorded point and line information_k(first view image I₁) The defect correction processing of (1). In addition, the present embodiment checks the k-th viewpoint image I in real time_k(first view image I₁) And performing a point defect judgment or a lineAnd judging defects.

Now, the defect correction according to the present embodiment (step S2 in fig. 17) will be explained. The present embodiment assumes the k-th viewpoint image I_k2j of the odd-numbered line_D-1 or even row 2j_DIs determined to have a line defect in the horizontal direction (x-direction), and the odd-numbered line 2j in the captured image I_D-1 or even row 2j_DIs not judged to have a line defect.

In the defect correction of step S2 of the present embodiment, the normal captured image I is set as the reference image, and the k-th viewpoint (first correction) image M is corrected based on the captured image I₁I_kThe defect of (2). The present embodiment is implemented by dividing the k-th viewpoint (first corrected) image M at a position not determined to be defective₁I_kIs compared with the signal value of the photographed image I at the position not determined as the defect to correct the defect. In this comparison, in order to perform defect correction with high precision, it is necessary to remove the k-th viewpoint image I caused by pupil shift_kAnd the respective shadow components of RGB of (1) and accurately comparing the k-th viewpoint images I_kAnd signal components of RGB of the subject between the captured images I. Thus, in step S1, the present embodiment corrects the shades (light amounts) of the RGB of the k-th viewpoint image in advance, generating a k-th viewpoint (first corrected) image M₁I_kThe same shadow state as the captured image I is established, and the influence of the shadow component is removed. Thereafter, in step S2, the present embodiment highly accurately corrects the shading-corrected kth-viewpoint (first correction) image M based on the captured image I₁I_kThe defect in (2) is corrected.

In step S2 of fig. 17, the present embodiment is based on the normal signal of the captured image I and the k-th viewpoint (first corrected) image M₁I_kTo the k-th viewpoint (first correction) image M after shading correction₁I_kAnd (j, i) performing defect correction processing on the signals judged to be defective. Then, the present embodiment generates a defect-corrected kth-viewpoint (second-corrected) image M₂I_k(j, i). Here, R, Gr, Gb are as followsAnd B each represents a k-th viewpoint (second corrected) image M employing a Bayer array₂I_k. In other words, the k-th viewpoint (second correction) image of R is represented as RM₂I_k(2j₂-1,2i₂-1)＝M₂I_k(2j₂-1,2i₂-1), and represents the kth viewpoint (second correction) image of Gr as GrM₂I_k(2j₂-1,2i₂)＝M₂I_k(2j₂-1,2i₂). Representing the kth viewpoint (second correction) image of Gb as GbM₂I_k(2j₂,2i₂-1)＝M₂I_k(2j₂,2i₂-1), and represents the k-th viewpoint (second corrected) image of B as BM₂I_k(2j₂,2i₂)＝M₂I_k(2j₂,2i₂)。

In step S2, assume that the kth viewpoint (first corrected) image M₁I_kFirst position (2 j) of R of (1)_D-1,2i_D-1) is judged as defective. At this time, based on the captured image RI (2 j) at the first position_D-1,2i_D-1), the k-th viewpoint (first corrected) image RM at the second position of R that is not judged to be defective₁I_kAnd a captured image RI at the second position, the defect correction processing is performed by the following expression (8A). This configuration generates a defect-corrected kth-viewpoint (second-corrected) image RM at the first position₂I_k(2j_D-1,2i_D-1)。

Assume that the kth viewpoint (first corrected) image M₁I_kFirst position (2 j) of Gr_D-1,2i_D) Is judged to be defective. At this time, based on the captured image GrI (2 j) at the first position_D-1,2i_D) And a k-th viewpoint (first corrected) image GbM at a second position of the Gb that is not determined to be defective₁I_kAnd a captured image GbI at the second position, defect correction processing is performed by the following expression (8B). This configuration generates a defect-corrected kth-viewpoint (second-corrected) image GrM at the first position₂I_k(2j_D-1,2i_D)。

Assume that the kth viewpoint (first corrected) image M₁I_kFirst position (2 j) of Gb of (1)_D,2i_D-1) is judged as defective. At this time, based on the captured image GbI (2 j) at the first position_D,2i_D-1), a k-th viewpoint (first corrected) image GrM at a second position of Gr not judged to be defective₁I_kAnd a captured image GrI at the second position, defect correction processing is performed by the following expression (8C). This configuration generates a defect-corrected kth-viewpoint (second-corrected) image GbM at the first position₂I_k(2j_D,2i_D-1)。

Assume that the kth viewpoint (first corrected) image M₁I_kFirst position (2 j) of B of (1)_D,2i_D) Is judged to be defective. At this time, based on the photographed image BI (2 j) at the first position_D,2i_D) And a k-th viewpoint (first corrected) image BM at a second position of B that is not determined to be defective₁I_kAnd the captured image BI at the second position, the defect correction processing is performed by the following expression (8D). This configuration generates a k-th viewpoint (second corrected) image BM after defect correction at the first position₂I_k(2j_D,2i_D)。

Expression 8A

Expression 8B

Expression 8C

Expression 8D

At most of the positions (j, i) not judged as defective, the k-th viewpoint (second corrected) image M₂I_k(j, i) has a (first corrected) image M corresponding to the k-th viewpoint₁I_k(j, i) has the same signal value, and M₂I_k(j,i)＝M₁I_k(j, i) holds. The k-th viewpoint (second corrected) image M after defect correction may be subjected to correction as needed₂I_k(j, i) as an output image.

Now, the first-viewpoint (first-correction) image M based on the image shown in step S2 of fig. 17 in the present embodiment will be described with reference to fig. 24 and 25₁I₁The normal captured image I. Fig. 24 shows the first-viewpoint (first-correction) image M before defect correction in the present embodiment₁I₁(after shading correction and demosaicing). Showing the image M at a first viewpoint (first correction)₁I₁An example in which a line defect occurs in the horizontal direction (x direction) in the central portion of (j, i). Fig. 25 shows the first-viewpoint (second-correction) image M after defect correction in the present embodiment₂I₁(after shading correction and demosaicing). The line defect in the horizontal direction (x direction) is corrected with the defect correction based on the normal captured image I, and the defect-corrected first-viewpoint (second-corrected) image M having good quality is generated as in the captured image₂I₁(j,i)。

The present embodiment generates a captured image corresponding to a pupil area obtained by combining different partial pupil areas based on an input image acquired by an image pickup element including a plurality of pixels having a plurality of photoelectric converters configured to receive light beams having passed through the different partial pupil areas in an imaging optical system. Then, the present embodiment generates a plurality of viewpoint images for respective different partial pupil areas, performs image processing for correcting the viewpoint images based on the captured images, and generates an output image. The present embodiment performs image processing based on a captured image to correct and reduce defects contained in a viewpoint image. The present embodiment performs image correction processing for correcting the signal value of the viewpoint image at the first position determined to be defective based on the signal value of the captured image at the first position. The present embodiment performs signal processing for correcting a signal value of a viewpoint image at a first position based on a signal value of a captured image at a first position determined to be defective, a signal value of a viewpoint image at a second position not determined to be defective, and a signal value of a captured image at the second position.

In the present embodiment, the image processing unit 125b performs correction processing (image processing) based on the captured image after performing light amount correction processing for the viewpoint image based on the captured image to reduce defects contained in the viewpoint image. This configuration can generate a viewpoint image with good quality.

Next, in step S2 of fig. 17, the image processing unit 125b corrects the k-th viewpoint (second correction) image M after the defect correction by using the following expressions (9A) to (9D)₂I_k(j, i) shading. Thereby, a k-th viewpoint (third corrected) image M is generated₃I_k(j,i)。

Expression 9ARM₃I_k(2j₂-1,2i₂-1)＝RSF_k(2i₂-1)×RM₂I_k(2j₂-1,2i₂-1),…(9A)

Expression 9BG_rM₃I_k(2j₂-1,2i₂)＝G_rSF_k(2i₂)×G_rM₂I_k(2j₂-1,2i₂),…(9B)

Expression 9CG_bM₃I_k(2j₂,2i₂-1)＝G_bSF_k(2i₂-1)×G_bM₂I_k(2j₂,2i₂-1),…(9C)

Expression 9DBM₃I_k(2j₂,2i₂)＝BSF_k(2i₂)×BM₂I_k(2j₂,2i₂)…(9D)

Now, a k-th viewpoint (third corrected) image M employing a bayer array is acquired for each of R, Gr, Gb, and B₃I_k. Expressing the k-th viewpoint (third corrected) image of R as RM₃I_k(2j₂-1,2i₂-1)＝M₃I_k(2j₂-1,2i₂-1), and represents the kth viewpoint (third corrected) image of Gr as GrM₃I_k(2j₂-1,2i₂)＝M₃I_k(2j₂-1,2i₂). Representing the kth viewpoint (third correction) image of Gb as GbM₃I_k(2j₂,2i₂-1)＝M₃I_k(2j₂,2i₂-1), and represents the k-th viewpoint (third corrected) image of B as BM₃I_k(2j₂,2i₂)＝M₃I_k(2j₂,2i₂)。

In step S3 of fig. 18, the captured image I (j, I) and the k-th viewpoint (third corrected) image M are subjected to₃I_k(j, i) performing saturation signal processing. This example discusses that k is 1 and N_LFExample 2.

In step S3, first, for a captured image I (j, I) in which the maximum value of the captured signal is set to Imax, saturation signal processing is performed using the following expression (10), and a corrected captured image MI (j, I) is generated.

Expression 10

Next, in step S3, the image processing unit 125b targets the k-th viewpoint (third corrected) image M₃I_k(j, i) (wherein the shading function of the Bayer array is set to SF_k(j, i)), the saturation signal processing corresponding to the shadow state is performed as in the following expression (11). Thereby, the kth viewpoint (fourth corrected) image M can be generated₄I_k(j, i). Here, the shading function RSF is generated based on each of R, Gr, Gb, and B_k(2i₂-1)、GrSF_k(2i₂)、GbSF_k(2i₂-1) and BSF_k(2i₂) The shading function SF of the Bayer array is calculated by expressions (6A) to (6D)_k(j, i). In other words, SF is assumed_k(2j₂-1,2i₂-1)＝RSF_k(2i₂-1)、SF_k(2j₂-1,2i₂)＝GrSF_k(2i₂)、SF_k(2j₂,2i₂-1)＝GbSF_k(2i₂-1) and SF_k(2j₂,2i₂)＝BSF_k(2i₂)。

Expression 11

In step S4 of fig. 18, the image processing unit 125b passes through the corrected photographed image MI (j, i) and the first-viewpoint (fourth-corrected) image M₄I₁(j, I) generating the second view image I based on the expression (12)₂(j,i)。

Expression 12

I₂(j,i)＝MI(j,i)-M₄I₁(j,i). (12)

In the present embodiment, the first-viewpoint (third-corrected) image M is due to the driving method of the image pickup element 107 and the a/D conversion circuit configuration₃I₁The maximum signal value at the saturation of (j, I) may have the same maximum signal value as the maximum signal value Imax at the saturation of the captured image I (j, I). In this case, the second viewpoint image is generated by subtracting the first viewpoint (third correction) image from the captured image as in expression 12 in a state where the saturation signal processing is not performed, and the second viewpoint imageAs may be the case with a saturated signal value, the saturated signal value may have the wrong signal value 0. Thus, to prevent this problem, step S3 is directed to the captured image I (j, I) and the k-th viewpoint (third corrected) image M₃I_k(j, i) saturation signal processing corresponding to the shading state is performed in advance so that the corrected captured image MI (j, i) and the first-viewpoint (fourth-corrected) image M subjected to the saturation signal processing are generated₄I₁(j, i). Thereafter, step S4 may generate the second view image I by using expression (12)₂(j, I) generating a second view image I corresponding to a correct saturation signal value₂。

In step S5 of fig. 18, the image processing unit 125b aims at the first-viewpoint (fourth-corrected) image M₄I₁(j, I) and a second view image I₂(j, i) shading correction (light amount correction) is performed. More specifically, the image M for the first viewpoint (fourth correction)₄I₁(j, i) using the shading function RSF which has been generated by the expressions (6A) to (6D)₁、GrSF₁、GbSF₁And BSF₁Shading correction (light amount correction) is performed in the same manner as expressions (7A) to (7D). Thereby, the first-viewpoint (fifth corrected) image M is generated₅I₁(j, i). Next, in step S5, the image processing unit 125b targets the second viewpoint image I₂(j, i), shading correction (light amount correction) is performed based on the corrected captured image MI (j, i) as in expressions (4A) to (7D). Thereby, the second-viewpoint (fifth corrected) image M is generated₅I₂(j,i)。

Finally, in step S6 of fig. 18, the image processing unit 125b aims at the first-viewpoint (fifth-corrected) image M₅I₁(j, i) and a second-viewpoint (fifth-corrected) image M₅I₂(j, i) the saturated signal processing is performed by the following expression (13). Thereby, the first corrected viewpoint image MI is generated₁(j, i) and a second corrected viewpoint image MI₂(j, i) as an output image.

Expression 13

Now, the second viewpoint image I shown in step S5 of fig. 17 will be explained with reference to fig. 26 and 27₂And (j, i) shading correction processing (light amount correction processing) for each of the RGB. FIG. 26 shows the second-viewpoint image I before shading correction in the present embodiment₂(after demosaicing) example. Pupil shift between the exit pupil of the imaging optical system and the entrance pupil of the image pickup element causes RGB to each generate a shadow, and thus an image I at the second viewpoint₂The reduction in luminance and the modulation of the RGB ratio occur on the left side of (j, i). FIG. 27 shows the shading-corrected second corrected viewpoint image MI in the present embodiment₂(after demosaicing) example. The reduction in luminance and the modulation of the RGB ratio are corrected based on the shading correction of each of RGB of the captured image, and the second corrected viewpoint image MI after shading correction having good quality is generated as in the captured image₂(j,i)。

The image processing apparatus according to the present embodiment is an image processing apparatus having an image processing unit for performing the above-described image processing method. An image pickup apparatus according to the present embodiment is an image pickup apparatus including an image processing unit for performing the above-described image processing method and an image pickup element including a plurality of arrayed pixels having a plurality of sub-pixels configured to receive light beams having passed through different partial pupil regions in an imaging optical system. The structure of the present embodiment can generate a viewpoint image with good quality.

Fourth embodiment

Next, a fourth embodiment according to the present invention will be explained. The present embodiment is based on the first corrected viewpoint image and the second corrected viewpoint image (first corrected viewpoint image to nth corrected viewpoint image) generated in the third embodiment_LFCorrected viewpoint image) and the correlation (degree of coincidence of signals) between the first corrected viewpoint image and the second corrected viewpoint image, an image shift amount distribution is detected by a phase difference detection method.

In generating an imageWhen the shift amount is distributed, first, the viewpoint image MI is corrected based on the k-th correction viewpoint image MI which is an RGB signal using a bayer array_k(k＝1～N_LF) For each position (j, i), the color barycenters for RGB are made to coincide with each other. The k-th viewpoint luminance signal Y is generated by the following expression (14)_k。

Expression 14

Next, in generating the image shift amount distribution, the first corrected viewpoint image MI based on the RGB signals as the signals using the bayer array₁For the first-viewpoint luminance signal Y generated by expression (14)₁A one-dimensional band-pass filter process is performed in the pupil division direction (column direction), and a first focus detection signal dYA is generated. In addition, the present embodiment is based on the second corrected viewpoint image MI₂For the second viewpoint luminance signal Y generated by expression (14)₂A one-dimensional band-pass filter process is performed in the pupil division direction (column direction), and a second focus detection signal dYB is generated. One-dimensional bandpass filters, for example, may use first order differential filters [1,5,8,8,8, 5,1, -1, -5, -8, -8, -8, -8, -5, -1]And the like. The pass band of the one-dimensional band-pass filter can be adjusted as desired.

Next, in generating the image shift amount distribution, the present embodiment shifts the first focus detection signal dYA and the second focus detection signal dYB relative to each other in the pupil division direction (column direction), calculates a correlation amount indicating the degree of coincidence of the signals, and generates the image shift amount distribution M based on the correlation amount_DIS(j, i). The present embodiment sets the first focus detection signal to dYA (j + j)₂,i+i₂) And the second focus detection signal is set to dYB (j + j)₂,i+i₂) Wherein, dYA (j + j)₂,i+i₂) J' th in the row direction with the position (j, i) as the center₂Root of Chinese character (-n)₂≤j₂≤n₂) And the ith in the column direction as the pupil division direction₂Root of large (-m)₂≤i₂≤m₂). Assume that the offset is s (-n)_s≤s≤n_s). Then, a correlation amount COR at each position (j, i) is calculated by the following expression (15A)_EVEN(j, i, s), and the correlation amount COR is calculated by the following expression (15B)_ODD(j,i,s)。

Expression 15A

Expression 15B

Correlation quantity COR_ODD(j, i, s) is determined by offsetting the amount of shift between the first focus detection signal dYA and the second focus detection signal dYB relative to the correlation amount COR_EVEN(j, i, s) is offset by half the phase or-1. The embodiment is based on the correlation COR_EVEN(j, i, s) and a correlation COR_ODD(j, i, s) calculating and averaging the shift amount of the real value minimizing the correlation amount by sub-pixel calculation, and detecting the image shift amount distribution M_DIS(j,i)。

In detecting the image shift amount by the phase difference method, the present embodiment evaluates the correlation amounts of expressions (15A) and (15B), and detects the image shift amount based on the correlation (the degree of coincidence of signals) between the first focus detection signal and the second focus detection signal. The present embodiment generates the first focus detection signal and the second focus detection signal from the first corrected viewpoint image and the second corrected viewpoint image after shading (light amount) correction is performed on each of RGB based on the captured image. Thus, the present embodiment can improve the correlation (the degree of coincidence of signals) between the first focus detection signal and the second focus detection signal, and detect the amount of image shift with high accuracy.

In the defocus amount detected from the auto focus detectionWhen the lens is driven to the focusing position, the image shift amount distribution M is_DIS(j, i) is multiplied by a conversion coefficient K from an image shift amount corresponding to lens information such as an aperture value F and an exit pupil distance of an imaging lens (imaging optical system) to a defocus amount. Thereby, the defocus distribution M can be detected_Def(j, i). This calculation may be performed for each image height position in the focus detection area.

The present embodiment can generate a plurality of viewpoint images with good quality. The present embodiment can improve the detection accuracy of the image shift amount by using a plurality of viewpoint images with good quality.

Fifth embodiment

Next, a fifth embodiment according to the present invention will be explained. This example discusses N_x＝2、N _y2 and N_LFAn example of a four split of 4. In the present embodiment, sub-pixels 201 to 204 (N) from four divisions are set for each pixel based on an input image (LF data) corresponding to the pixel array shown in fig. 12 and expression (2) (N)_x×N_yDivided first to Nth sub-pixels_LFSub-pixels) are synthesized. Generating a captured image I as having a pixel number N (horizontal pixel number N)_HX number of vertical pixels N_V) RGB signals of a bayer array of resolution.

This example discusses N_x＝2、N _y2 and N_LFFour divisions of 4 and k 1-3. The signal of the sub-pixel 201 (first sub-pixel) is selected from the four-divided sub-pixels 201 to 204 for each pixel based on LF data (input image) corresponding to the pixel array shown in fig. 12 and expression (3). Then, the present embodiment generates the first-viewpoint image I₁(j, i) RGB signals as a bayer array having a resolution of the number of pixels N corresponding to the partial pupil area 501 in the imaging optical system. The present embodiment selects a signal of the sub-pixel 202 (second sub-pixel) from the four-divided sub-pixels 201 to 204 for each pixel based on LF data and expression (3). The present embodiment generates the second viewpoint image I₂(j, i) as having an andRGB signals of a bayer array of resolution of the number of pixels N corresponding to the partial pupil area 502 in the image optical system. The present embodiment selects a signal of the sub-pixel 203 (third sub-pixel) from the four-divided sub-pixels 201 to 204 for each pixel based on LF data and expression (3). The present embodiment generates the third viewpoint image I₃(j, i) RGB signals as a bayer array having a resolution of the number of pixels N corresponding to the partial pupil area 503 in the imaging optical system.

The present embodiment generates a captured image corresponding to a pupil area obtained by combining different partial pupil areas based on an input image (LF data) acquired by an image pickup element including a plurality of arrayed pixels having a plurality of photoelectric converters configured to receive light beams having passed through the different partial pupil areas in an imaging optical system. Then, the present embodiment generates a plurality of viewpoint images for different partial pupil areas, respectively. In order to generate each viewpoint image with good quality, the present embodiment performs the first to fourth viewpoint images (first to nth viewpoint images) based on the captured image, as in the third embodiment_LFViewpoint image) performs image processing such as stain correction and shading correction, and generates an output image.

In step S1 of fig. 17, the present embodiment sets the captured image I (j, I) as a reference or reference image for the first viewpoint image I₁To the third viewpoint image I₃(kth view image I_k:k＝1～N_LF-1) shading (or light amount) correction. This example discusses N_x＝2、N _y2 and N_LFFour divisions of 4 and k 1-3.

First, in step S1, the present embodiment is directed to the k-th viewpoint image I_k(k＝1～N_LF-1), shading (light amount) correction is performed in the x direction by expressions (4A) to (7D). Next, in expressions (4A) to (7D), the x direction is replaced with the y direction, and shading (light amount) correction processing is performed in the y direction, and the k-th viewpoint (first correction) image M is generated₁I_k(k＝1～N_LF-1). In performing shading (light quantity) correction in the x-direction and the y-directionIn the case of two stages of shading correction in the upward direction, the pupil division number N _LF1 more than the required number so that expressions (5A) to (5D) are normalized. Thus, at the time of the second shading correction in the y direction, the sum of the pupil division number N is omitted in expressions (5A) to (5D)_LFMultiplied together for normalization.

By expressions (8A) to (10), the procedure of the present embodiment is the same as that of the third embodiment until the kth viewpoint (fourth correction) image M is generated₄I_k(k＝1～N_LF-1) up to. In step S4 of fig. 18, the corrected photographed image MI (j, i) and the k-th viewpoint (fourth corrected) image M are used as the basis₄I_k(k＝1～N_LF-1) generating the Nth by the following expression (16)_LFViewpoint image I_NLF(j, i). This example discusses N_x＝2、N _y2 and N_LFAn example of a four split of 4.

Expression 16

Step S5 and subsequent steps in fig. 18 are the same as those in the third embodiment.

The present embodiment can generate a viewpoint image with good quality. In the photoelectric converter of each pixel of the image pickup element, other embodiments may increase the division number, such as N_x＝3、N_y3 and N_LFNine divisions of 9 and N_x＝4、N_y4 and N_LFSixteen divisions of 16, etc.

As described above, the image processing apparatus (image processing circuit 125) in each embodiment includes the acquisition unit 125a and the image processing unit 125b (correction unit). The acquisition unit 125a acquires a parallax image generated based on a signal of one of a plurality of photoelectric converters for receiving light beams passing through mutually different partial pupil areas in the imaging optical system, and acquires a captured image generated by synthesizing the signals from the plurality of photoelectric converters. The image processing unit 125b performs correction processing based on the captured image so that defects (such as point flaws and line flaws) contained in the parallax image are reduced.

Preferably, the image processing unit 125b corrects the pixel value (pixel signal) of the parallax image at the first position by using the pixel value (pixel signal) of the captured image at the first position (position to be corrected) determined as the defect. More preferably, the image processing unit 125b corrects the pixel value of the parallax image at the first position based on the pixel value I of the captured image at the first position, the pixel value of the parallax image at the second position which is not determined to be defective, and the pixel value of the captured image at the second position. The second position is a position of a pixel near (or in the periphery of) the first position (position to be corrected). More preferably, the second position is a position adjacent to the first position in a predetermined direction (a vertical direction, a horizontal direction, or an oblique direction on the pixel array).

Preferably, in the case where the pixel value of the parallax image or the captured image at the second position is lower than a predetermined luminance value (parameter a0 or I0), the image processing unit 125b replaces the pixel value with the predetermined luminance value. More preferably, the predetermined luminance value is set to change in accordance with the number of partial pupil areas. Preferably, the predetermined luminance value is set to change in accordance with the first position (position to be corrected). Preferably, the predetermined luminance value is set to change according to the image capturing condition information. The imaging condition information includes at least one of an ISO sensitivity, an aperture value of the imaging optical system, and an exit pupil distance.

Preferably, the image processing apparatus includes a memory (memory 134) for storing the defect information about the first location, or an inspector for inspecting the defect information about the first location. The image processing unit 125b performs correction processing based on the defect information stored in the memory or the defect information obtained as the inspection result of the inspector.

Preferably, the parallax image is generated by collecting light reception signals of a plurality of sub-pixels (a plurality of first sub-pixels, a plurality of second sub-pixels, a plurality of third sub-pixels, and a plurality of fourth sub-pixels) included in one photoelectric converter for respective partial pupil regions different from each other in the imaging optical system. The captured image is generated by collecting light reception signals of all the sub-pixels (the plurality of first sub-pixels and the plurality of second sub-pixels, and further, the plurality of third sub-pixels and the plurality of fourth sub-pixels) included in the plurality of photoelectric converters.

Preferably, the parallax images include first and second parallax images corresponding to the respective light fluxes that are distinguished by passing through mutually different partial pupils in the imaging optical system. Then, the acquisition unit 125a acquires the captured image and the first parallax image of the parallax images from the image pickup element 107 including a plurality of photoelectric converters. The image processing unit 125b performs correction processing on the first parallax image in the parallax images. Then, the image processing unit 125b generates a second parallax image by subtracting the corrected first parallax image from the captured image. Preferably, the image processing unit 125b (refocus unit) performs refocus processing on the captured image based on the corrected parallax image.

Preferably, the image processing unit performs light amount correction or shading correction of the parallax image based on the captured image. More preferably, the image processing unit performs light amount correction processing of the parallax image for each color of the parallax image based on the captured image. More preferably, the image processing unit performs light amount correction processing of the parallax image based on the projection signal of the captured image and the projection signal of the parallax image. More preferably, after the light amount correction processing is performed on the parallax image by the image processing unit, the image processing unit corrects the parallax image based on the captured image so that defects included in the parallax image after the light amount correction processing is performed are reduced.

OTHER EMBODIMENTS

The present invention can be implemented in a process for supplying a program for implementing one or more functions of the above-described embodiments to a system or apparatus via a network or a storage medium, and reading and executing the program in one or more processors in a computer of the system or apparatus. In addition, the present invention may also be implemented in circuitry (such as an ASIC or the like) for performing one or more functions.

Embodiments may provide an image processing apparatus, an image capturing apparatus, an image processing method, a program, and a non-transitory computer-readable storage medium capable of generating a parallax image with improved quality.

Although the present invention has been described with reference to the exemplary embodiments, the present invention is not limited to the disclosed exemplary embodiments, and various changes and modifications may be made without departing from the scope of the present invention.

Claims

1. An image processing apparatus comprising:

an acquisition unit configured to acquire a parallax image generated based on a signal from one of a plurality of photoelectric converters corresponding to the same microlens of an image pickup element including a plurality of pixels provided with the plurality of photoelectric converters and each of the plurality of pixels corresponding to each of the plurality of microlenses, and acquire a composite image which is an image generated by synthesizing a plurality of signals from the plurality of photoelectric converters including the one photoelectric converter; and

an image processing unit for calculating a correction value by using a signal of the synthesized image as a reference image and correcting the parallax image using the correction value,

wherein the image processing unit corrects the pixel value of the parallax image at a first position based on the pixel value of the synthesized image at the first position, the pixel value of the parallax image at a second position, and the pixel value of the synthesized image at the second position.

2. The image processing apparatus according to claim 1, wherein the first position is a position determined to be defective, and the second position is a position not determined to be defective.

3. The image processing apparatus according to claim 1, wherein the second position is a position adjacent to the first position in a predetermined direction.

4. The image processing apparatus according to claim 1, wherein in a case where a pixel value of the parallax image or the synthesized image at the second position is lower than a predetermined luminance value, the image processing unit replaces the pixel value with the predetermined luminance value.

5. The image processing apparatus according to claim 4, wherein the predetermined luminance value is set to change in accordance with the number of partial pupil areas.

6. The image processing apparatus according to claim 4, wherein the predetermined luminance value is set to change in accordance with the first position.

7. The image processing apparatus according to claim 4, wherein the predetermined luminance value is set to change in accordance with image capturing condition information.

8. The image processing apparatus according to claim 7, wherein the imaging condition information includes at least one of an ISO sensitivity, an aperture value of an imaging optical system, and an exit pupil distance.

9. The image processing apparatus according to claim 1, further comprising a memory for storing defect information relating to the first position,

wherein the image processing unit corrects the parallax image based on the defect information.

10. The image processing apparatus according to claim 1, further comprising an inspector that inspects defect information relating to the first position,

11. The image processing apparatus according to any one of claims 1 to 10,

the parallax image is generated by collecting light reception signals from a plurality of sub-pixels included in the one photoelectric converter for respective partial pupil regions different from each other in the imaging optical system, an

The composite image is generated by collecting light reception signals from all the sub-pixels included in the plurality of photoelectric converters.

12. The image processing apparatus according to claim 1, wherein the parallax images include first and second parallax images corresponding to light fluxes passing through mutually different partial pupil regions in an imaging optical system, respectively,

wherein the acquisition unit acquires the composite image and the first parallax image of the parallax images from the image pickup element including the plurality of photoelectric converters, an

The image processing unit performs the following operations:

correcting the first parallax image of the parallax images, and

generating the second parallax image by subtracting the corrected first parallax image from the composite image.

13. The image processing apparatus according to claim 1, wherein the image processing unit performs refocus processing on the synthesized image based on the corrected parallax image.

14. The image processing apparatus according to claim 1, wherein the image processing unit performs light amount correction processing on the parallax image based on the synthesized image.

15. The image processing apparatus according to claim 14, wherein the image processing unit performs the light amount correction processing on the parallax image for each color of the parallax image based on the synthesized image.

16. The image processing apparatus according to claim 14, wherein the image processing unit performs the light amount correction processing on the parallax image based on a projection signal of the synthesized image and a projection signal of the parallax image.

17. The image processing apparatus according to any one of claims 14 to 16, wherein the image processing unit corrects the parallax image based on the synthesized image after the light amount correction processing on the parallax image is performed, to reduce a defect included in the parallax image after the light amount correction processing is performed.

18. An image pickup apparatus includes:

an image pickup element including a plurality of arrayed pixels, wherein the plurality of arrayed pixels are provided with a plurality of photoelectric converters, and each arrayed pixel of the plurality of arrayed pixels corresponds to each microlens of a plurality of microlenses;

an acquisition unit configured to acquire a parallax image generated based on a signal from one of the plurality of photoelectric converters corresponding to the same microlens of the image pickup element including a plurality of pixels provided with the plurality of photoelectric converters and acquire a composite image which is an image generated by synthesizing signals from the plurality of photoelectric converters including the one photoelectric converter; and

19. The image pickup apparatus according to claim 18, wherein the image pickup element includes the plurality of photoelectric converters for one microlens, and the microlenses are arranged in a two-dimensional manner.

20. An image processing method comprising the steps of:

acquiring a parallax image generated based on a signal from one of a plurality of photoelectric converters corresponding to the same microlens of an image pickup element including a plurality of pixels provided with the plurality of photoelectric converters and each of the plurality of pixels corresponding to each of the plurality of microlenses, and acquiring a composite image which is an image generated by synthesizing a plurality of signals from the plurality of photoelectric converters including the one photoelectric converter; and

by calculating a correction value using a signal of the synthesized image as a reference image and correcting the parallax image using the correction value,

wherein the pixel value of the parallax image at a first position is corrected based on the pixel value of the synthesized image at the first position, the pixel value of the parallax image at a second position, and the pixel value of the synthesized image at the second position.

21. A non-transitory computer-readable storage medium storing a program that causes a computer to execute a process comprising: