JP2018004706A - Focus detection device, focus detection method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a focus detection device capable of performing focus detection while reducing the influence of noise.SOLUTION: The focus detection device includes: first acquiring means (219) for performing a correlation operation while shifting a pair of parallax image signals in a first phase shift range and acquiring a correlation operation curve; replacing means (230) for replacing, of the correlation operation curve acquired by the first acquiring means, a correlation operation curve in a second phase shift range, which is a range sandwiching the position at which the pair of parallax image signals are not shifted and a range sandwiched between the minimum values having a high correlation of the pair of parallax image signals; and second acquiring means (222) for acquiring a defocus amount from the correlation operation curve whose a part is replaced by the replacing means.SELECTED DRAWING: Figure 16

Description

  The present invention relates to a focus detection apparatus, a focus detection method, and a program using a phase difference detection method.

  Conventionally, the luminous flux of the subject image is divided on the exit pupil plane, and each is imaged on each of a pair of pixel rows in a two-dimensional image sensor, and the phase difference of each image signal photoelectrically converted by the pair of pixel rows is calculated. A method of performing focus detection based on this is known as a phase difference detection type focus detection method.

  Patent Document 1 describes an imaging apparatus that performs phase difference detection type focus detection using a two-dimensional imaging element in which a microlens is formed in each pixel. In this imaging apparatus, each pixel has a photodiode A that photoelectrically converts one of the pupil-divided images, a photodiode B that photoelectrically converts the other image, and a floating diffusion that temporarily holds charges from the photodiodes A and B. Has a region. In the transfer control of the charges from the photodiodes A and B to the floating diffusion region, only the charge of the photodiode A is first transferred and converted into a voltage, whereby one image signal divided into pupils (hereinafter referred to as an A image signal). .) Can be obtained. Further, here, without resetting the floating diffusion region, the charge of the photodiode B is transferred to the floating diffusion region for voltage conversion, so that the image signal before pupil division (hereinafter referred to as A + B image signal). Can be obtained.

  An image signal corresponding to the photodiode B on which the other pupil-divided image is incident (hereinafter referred to as a B image signal) is obtained electrically by subtracting the A image signal from the A + B image signal. Since the image signal of the pair of A image signal and B image signal thus obtained is a parallax image signal, the phase difference of the A image signal and the B image signal is obtained by a known correlation calculation. The focus position of the subject can be detected. The A + B image signal is used to generate a captured image by the imaging device. As described above, the imaging apparatus described in Patent Document 1 can acquire three types of image signals, that is, an A image, a B image, and an A + B image, by two readings. In addition, since the A + B image signal is generated by mixing charges in the floating diffusion region, the A + B image signal is less noise than reading out the A image signal and the B image signal and adding the A image signal and the B image signal. Is obtained. This is because when the A image signal and the B image signal are read out, noise is included in each signal, and noise is also added by adding the signals.

JP 2014-89260 A

  In the circuit configuration of photoelectric conversion as described in Patent Document 1, light shot noise and dark current noise are mixed in the floating diffusion region. Therefore, the charge transferred from the photodiode is mixed with noise having different charge amounts depending on the timing temporarily held in the floating diffusion region. In particular, when photoelectrically converting a pair of pupil-divided images, the timing at which the A image signal is temporarily held in the floating diffusion region and the timing at which the A + B image signal is similarly temporarily held are also provided. , Noise of different charge amount is mixed. Therefore, the B image signal generated by subtracting the A image signal from the A + B image signal includes a signal component (referred to as inverse waveform noise) whose polarity is opposite to that of the noise component included in the A image signal.

  In the pair of pixel columns, the coordinates of the inverse waveform noise appearing in the B image signal are the same as the coordinates of the noise appearing in the A image signal. If there is no influence of noise, the amount of correlation becomes the smallest when the phase shift amount (phase difference) = 0 at the time of focusing. However, if the difference between the A + B image signal and the A image signal is thus set as the B image signal, the phase shift When the amount = 0, the influence of noise is larger than when other shift amounts are used. As a result, there is a phase difference that takes a smaller amount of correlation than when the phase shift amount = 0, and the phase difference may be recognized as the phase difference of the in-focus position, and erroneous ranging or hunting may occur. . Correlation at the time of calculation where the phase difference is zero deteriorates regardless of the focus detection state, so the correlation is particularly great when focus detection is performed in the vicinity of the in-focus position where the pixel row shows high correlation near the phase difference zero. As a result of deterioration, an error may occur in focus detection. In addition, when the subject is particularly low-luminance or low-contrast, the amplitude of the image signal caused by the original subject is small, so the influence of the noise is relatively high, and only when calculating the phase difference zero. In other words, the correlation deteriorates even when calculating the phase difference in the vicinity thereof.

  Further, the noise components included in the image signal pair have a correlation not only when the B image signal is acquired from the difference between the A + B image signal and the A image signal. For example, even when a noise source in the signal path is shared, such as when the A image signal and the B image signal are amplified by the same amplifier, the signal trains constituting the image signal pair (A image signal and B image signal) ) Have a correlation.

  Accordingly, the present invention has been made in view of the above-described problems, and one of its purposes is to provide a focus detection apparatus that can perform focus detection while reducing the influence of noise.

  To achieve the above object, the focus detection apparatus of the present invention includes a first acquisition unit that performs a correlation calculation while shifting a pair of parallax image signals within a first phase shift range, and acquires a correlation calculation curve; Of the correlation calculation curve of the first phase shift range acquired by the first acquisition means, the correlation between the pair of parallax image signals is high in a range sandwiching a position where the pair of parallax image signals is not shifted. The defocus amount is acquired from a replacement unit that replaces the correlation calculation curve in the second phase shift range that is between the minimum values, and the correlation calculation curve that is partially replaced by the replacement unit. A second acquisition unit.

  ADVANTAGE OF THE INVENTION According to this invention, the focus detection apparatus which can reduce the influence of noise and can perform focus detection can be provided.

1 is a diagram illustrating a schematic configuration of a digital camera according to an embodiment of the present invention. It is the schematic diagram which looked at the pixel surface of the image pick-up part from the incident light side. It is a figure which shows the circuit structure of the pixel of an imaging part. It is a drive timing chart in the circuit of a pixel. It is a figure which shows the structure of the pixel of an imaging part. It is a figure which shows the phase difference of the image signal of 1st subpixel S1 and 2nd subpixel S2 at the time of focusing. It is a figure which shows the phase difference of the image signal of 1st subpixel S1 and 2nd subpixel S2 at the time of a non-focusing. It is a figure which shows the relationship between the phase shift amount and defocus amount of a pair of image signal. FIG. 6 is a schematic diagram illustrating a subject image projected on a sub-pixel row when a defocus amount is small. It is a schematic diagram showing a subject image projected on a sub-pixel row when the defocus amount is large. This is a pattern subject used to explain noise appearing in an image signal. It is a graph (no noise mixing) showing the image signals of the first and second subpixel columns and the addition signal thereof for the pattern subject. It is the graph (with noise mixing) which showed the image signal of the 1st and 2nd subpixel row | line with respect to a pattern subject, and its addition signal. It is the graph which showed the correlation amount of the image signal pair in FIG. 13 for every phase shift amount. It is the graph which showed the correlation amount of the image signal pair in FIG. 14 for every phase shift amount. It is a flowchart which shows the flow of the focus detection operation | movement of the digital camera which concerns on embodiment of this invention. It is an image figure which shows the relationship between the correlation amount and phase shift amount after correlation amount substitution which concerns on embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described. Here, an embodiment in which the focus detection apparatus according to the present invention is applied to a lens interchangeable digital single-lens reflex camera will be described. However, the focus detection apparatus according to the present invention can be applied to any electronic device having an image sensor that can generate a signal used for phase difference detection focus detection. Examples of the electronic device include a digital camera of a type in which a lens cannot be exchanged, a video camera, a mobile phone equipped with a camera, a personal computer, a game machine, and a home appliance.

  The focus detection apparatus of the present embodiment is a focus detection apparatus that performs focus detection by an image plane phase difference method, and uses a pair of parallax image signals (A image signal and B image signal), and the B image signal with respect to the A image signal. A correlation calculation is performed by changing a phase shift amount (also referred to as a phase difference) to obtain a correlation calculation curve. Then, the phase difference having the minimum correlation amount (that is, the correlation is maximized) is searched, and the defocus amount is acquired based on the search result. The B image signal is obtained from the difference between the A + B image signal and the A image signal. At this time, the correlation amount acquired in a range susceptible to the influence of the inverse waveform noise is replaced with another value, and the phase difference at which the correlation amount takes the minimum value is searched using the correlation amount after the replacement. The replacement value is a value obtained by reducing the influence of the inverse waveform noise from the acquired correlation amount, whereby the focus detection can be performed while reducing the influence of the inverse waveform noise.

  Hereinafter, the present embodiment will be described in detail with reference to the drawings.

<Configuration of digital camera system>
FIG. 1 is a diagram illustrating a schematic configuration of a digital camera system which is an example of an imaging apparatus according to the present embodiment.

  In FIG. 1, the digital camera system includes a lens unit 100 and a camera unit 200. The lens unit 100 is detachably attached to the camera unit 200 via a lens mounting mechanism of a mount unit (not shown). An electrical contact unit 108 is provided on the mount portion. The electrical contact unit 108 has a communication bus line terminal including a communication clock line, a data transmission line, a data reception line, and the like. As a result, the lens unit 100 and the camera unit 200 can communicate with each other.

  The lens unit 100 includes a lens group 101 that forms an imaging optical system and includes a focus lens and a zoom lens, and a diaphragm 102 that controls incident light. In addition, a drive system (not shown) including a stepping motor for performing zoom and focus of the lens group 101 and a lens drive unit 103 for controlling the drive system are provided. The lens driving unit 103 and the driving system constitute a focus adjusting unit.

  In addition, the lens unit 100 includes an aperture control unit 104 that controls the aperture of the aperture 102 and an optical information recording unit 106 that records various optical design values of the zoom, focus, and aperture of the lens group 101.

  The lens driving unit 103, the aperture control unit 104, and the optical information recording unit 106 are connected to a lens controller 105 that includes a CPU that controls the operation of the entire lens unit 100.

  In addition, the lens unit 100 is provided with a lens position detection unit 107 that acquires the phase waveform of the stepping motor included in the lens driving unit 103 from the lens driving unit 103 through the lens controller 105 and detects lens position information. ing.

  The camera unit 200 communicates with the lens unit 100 via the electrical contact unit 108, transmits a control request for the zoom and focus of the lens group 101, and the aperture of the diaphragm 102, and receives a control result.

  The incident light beam is guided through the lens group 101 and the diaphragm 102 to an image capturing unit 213 that includes an image sensor (not shown) and a processor that performs a development operation in the camera unit 200. In the imaging unit 213, imaging data is obtained by photoelectrically converting an incident subject image and performing development calculation.

  In addition, the camera unit 200 is provided with an operation switch 214 for performing an operation input to the camera unit 200. The operation switch 214 is a two-stage stroke type switch. The first-stage switch (SW1) is a switch for starting a shooting preparation operation such as photometry or focus detection using an imaging signal. The second-stage switch (SW2) is a switch for starting an imaging operation such as charge accumulation and charge reading in the imaging unit 213 in order to acquire a still image.

  The imaging unit 213 includes an imaging device having a plurality of pixels each including a photoelectric conversion unit, and an A / D converter that converts an electrical signal that is an analog signal output from the imaging device into an image signal that is digital data. As the image sensor, for example, a CCD image sensor or a CMOS image sensor can be used. Further, the A / D converter may be incorporated in the image sensor. In this specification, a pixel may output an electric signal corresponding to an image signal, and a pixel may simply output an image signal. Of the pixels included in the image sensor, at least some of the pixels can output an image signal used for focus detection. An image signal for focus detection obtained by photoelectric conversion at the pixel is temporarily stored in a memory 216 connected to the camera controller 215. A pair of parallax image signals (hereinafter simply referred to as a pair of image signals) temporarily stored in the memory 216 is sent to the pixel addition unit 217 connected to the camera controller 215.

  The pixel addition unit 217 adds the image signal obtained from the positionally corresponding pixel to the pair of parallax image signals until the number of times counted by the addition counter 218 reaches a predetermined number. The added pair of image signals is sent to a correlation calculation unit 219 which is a correlation amount acquisition unit connected to the camera controller 215, and the correlation calculation unit 219 performs correlation for each phase shift amount of the pair of image signals. Get the quantity and get the correlation calculation curve. This correlation calculation is performed in the first phase shift range. The acquired correlation amount is subjected to replacement of a part of the correlation amount (correlation amount in the second phase shift range) by a correlation amount interpolation unit 230 which is a replacement unit for replacing the correlation amount under a predetermined condition described later. The detailed operation of the correlation amount interpolation unit will be described later in detail. In order to acquire the correlation amount after replacement in the manner of interpolating the correlation amount in the range in which the correlation amount is replaced from the correlation amount in the range in which the correlation amount is not replaced. In this embodiment, the replacement means is referred to as a correlation amount interpolation unit 230. The correlation amount output from the correlation amount interpolation unit 230 is added by the correlation amount addition unit 220 connected to the camera controller 215 until the number of times counted by the addition counter 218 reaches a predetermined number. The added correlation amount is sent to the phase difference detection unit 221 to obtain the phase difference having the highest correlation amount (phase difference indicating the minimum correlation amount).

  A defocus amount detection unit 222 that is a defocus amount acquisition unit acquires a defocus amount by a known phase difference detection method based on the phase difference acquired by the phase difference detection unit 221 and the optical characteristics of the lens unit 100. . The calculated defocus amount is sent to the defocus amount addition unit 223, and is added until the number of times counted by the addition counter 218 reaches a predetermined number.

  The correlation amount acquisition unit may be referred to as a first acquisition unit, and the defocus amount acquisition unit may be referred to as a second acquisition unit.

  The camera controller 215 transmits / receives control information to / from the lens controller 105 via the electrical contact unit 108, and based on the defocus amount calculated by the defocus amount detection unit 222 or the defocus amount addition unit 223, Control the focal position.

  The digital camera according to the present embodiment includes a display unit 224 that displays a subject image captured by the imaging unit 213 and various operation situations, and two operation modes, a live view mode or a moving image recording mode, which will be described later. An operation unit 225 for switching to is provided. Further, in the digital camera, the live view mode is automatically set when the power is turned on, the imaging unit 213 starts imaging, the captured subject image is developed and displayed on the display unit 224. Still images and moving images captured by the imaging unit 213 are recorded in the recording unit 226 in a predetermined data format.

  In addition, the digital camera uses a temperature measurement unit 227 that measures the temperature of the imaging unit 213 and an image signal from the imaging unit 213 to generate a parallax image signal pair that is a target of correlation calculation by the correlation calculation unit 219. And processing means. The digital camera of the present embodiment has, as image processing means, a pixel subtraction unit 228 that inputs two image signals from the imaging unit 213 and subtracts the other from one image signal, and one-dimensional low-pass filtering ( LPF processing unit 229 for performing (LPF).

  Next, the configuration of the image sensor will be described with reference to FIGS.

  FIG. 2 is a schematic view of the light receiving surface of the image sensor as viewed from the incident light side. Each of the pixels included in the imaging device includes a first sub-pixel S1 and a second sub-pixel S2 into which a pair of light beams divided on the exit pupil plane respectively enter. A light beam passing through the first pupil region of the exit pupil is incident on the first subpixel S1, and a light beam passing through the second pupil region of the exit pupil is incident on the second subpixel S2. Microlenses ML are disposed on the front surfaces of the first subpixel S1 and the second subpixel S2 for condensing light. On the light receiving surface of the image sensor, pixels each including the first sub-pixel S1, the second sub-pixel S2, and the micro lens ML are arranged in the horizontal direction as h pixels and in the vertical direction as v pixels. The image signals from the first and second subpixels S1 and S2 can be used for focus detection, and the image signal from the unit pixel can be used for imaging. In addition, the first subpixel and the second subpixel that constitute the same pixel may be referred to as a subpixel that forms a pair. Light beams that have passed through the same microlens ML are incident on the sub-pixels forming a pair.

  FIG. 3 is a diagram illustrating a circuit configuration of a pixel. The pixel 300 includes a first photodiode 301A, a second photodiode 301B, a first transfer switch 302A, a second transfer switch 302B, and a floating diffusion region 303. Further, the pixel 300 includes an amplifying unit 304, a reset switch 305, and a selection switch 306. Reference numeral 308 denotes a common power supply VDD. The first photodiode 301A and the first transfer switch 302A constitute the first subpixel S1, and the second photodiode 301B and the second transfer switch 302B constitute the second subpixel S2.

  The first and second photodiodes 301A and 301B photoelectrically convert light rays that pass through the same microlens ML and are incident on each. The first and second transfer switches 302A and 302B control to selectively transfer charges generated in the first and second photodiodes 301A and 301B to the common floating diffusion region 303, respectively. The first and second transfer switches 302A and 302B are controlled by first and second transfer pulse signals PTXA and PTXB, respectively. The floating diffusion region 303 temporarily holds the charges transferred from the first and second photodiodes 301A and 301B, respectively. The floating diffusion region 303 has an effect of converting the retained charge into a voltage signal. The amplifying unit 304 is composed of a source follower MOS transistor, amplifies a voltage signal based on the charge held in the floating diffusion region 303, and outputs it as a pixel signal. The reset switch 305 resets the potential of the floating diffusion region 303 to the reference potential VDD. The reset switch 305 is controlled by a reset pulse signal PRES. The selection switch 306 controls the output of the pixel signal from the amplification unit 304 to the vertical output line 307. The selection switch 306 is controlled by a vertical selection pulse signal PSEL.

  FIG. 4 is a timing chart showing driving of the circuit of the pixel 300.

  First, between time t1 and t2, the first and second photodiodes 301A and 301B are set by setting the first and second transfer pulse signals PTXA and PTXB to High while the reset pulse signal PRES is High. To reset.

  At time t2, the first and second transfer pulse signals PTXA and PTXB are set to Low to start accumulation of electric charges in the first and second photodiodes 301A and 301B.

  After the charge is accumulated for a necessary time based on the desired exposure amount, the vertical selection pulse signal PSEL is set to High at time t3 to turn on the selection switch 306. Subsequently, at time t4, the reset pulse signal PRES is set to Low to release the reset of the floating diffusion region 303.

  Between time t5 and time t6, the first transfer pulse signal PTXA is set to High, thereby turning on the first transfer switch 302A and transferring the charge of the first photodiode 301A to the floating diffusion region 303.

  At time t6, the first transfer pulse signal PTXA is set to Low, so that the charge accumulated in the first photodiode 301A is read to the floating diffusion region 303. By reading the charge, a voltage signal corresponding to the change in potential is output to the vertical output line 307 as the first focus detection pixel signal via the amplifier 304 and the selection switch 306.

  Between time t7 and time t8, the first and second transfer switches 302A and 302B are turned ON by setting the first transfer pulse signal PTXA to High and simultaneously setting the second transfer pulse signal PTXB to High. As a result, the charges of the first and second photodiodes 301 </ b> A and 301 </ b> B are simultaneously transferred to the floating diffusion region 303.

  At time t8, the first and second transfer pulse signals PTXA and PTXB are set to Low, so that the transferred charges are read to the floating diffusion region 303. By reading the charge, a voltage signal corresponding to the change in potential is output to the vertical output line 307 as an addition signal of the first and second focus detection pixel signals through the amplifier 304 and the selection switch 306.

  At time t9, the reset pulse signal PRES is set to High to reset the floating diffusion region 303.

  The above operation is sequentially performed for each unit pixel, and an image signal acquired from the pixel column of the first subpixel and an addition signal of the image signal acquired from the pixel column of the first and second subpixels are read out. Is completed. In this specification, an image signal acquired from the pixel row of the first subpixel is an A image signal, and an image signal acquired from the pixel row of the second subpixel is a B image signal. An image signal acquired from the pixel row of the second sub-pixel is assumed to be an A + B image signal.

  The read A image signal and A + B image signal are input to the pixel subtracting unit 228, and by taking a difference between these image signals, an image signal (B image signal) of the pixel column of the second sub-pixel is electrically obtained. ) Is generated. In the present invention and the present specification, as in the present embodiment, even when the B image signal is indirectly acquired without actually reading out only the B image signal, It is assumed that it is obtained from the second subpixel or the second photodiode 301B.

  FIG. 5 is a cross-sectional view showing the structure of the unit pixel in FIG. In FIG. 5, a microlens ML is disposed on the light incident side of the first subpixel S1 and the second subpixel S2. Reference numeral 501 denotes a smooth layer that forms a plane on which the microlenses ML are arranged. Downstream of the light shielding layer 502, the first and second photodiodes 301A and 301B are arranged, and the light beam passing through the first pupil region of the exit pupil is directed to the first photodiode 301A. The light beam that passes through the second pupil region enters the second photodiode 301B. When the unit pixel has such a configuration, the pupil of the imaging optical system is divided symmetrically, and image signals corresponding to the divided light beams can be acquired.

<Focus detection using image signals acquired from the pixel rows of the first and second sub-pixels>
In FIG. 2, an image signal acquired from the first subpixel column and the second subpixel column that form a pair includes two images approximated as the number of subpixels constituting the subpixel column increases. Become. When the imaging optical system is in focus with respect to the subject, the A image signal output from the pixel column of the first subpixel S1 and the B image signal output from the pixel column of the second subpixel S2 Are almost identical to each other. On the other hand, if the imaging optical system is out of focus, a phase difference occurs between the A image signal and the B image signal. The direction of the phase difference is reversed between the front pin state and the rear pin state.

  FIG. 6 is a diagram illustrating the phase difference between the image signals of the first subpixel S1 and the second subpixel S2 when in focus. On the other hand, FIG. 7 is a diagram showing the phase difference between the image signals of the first subpixel S1 and the second subpixel S2 when the image is out of focus.

A light flux from a specific point on the subject is a light flux ΦLa that enters the subpixel S1 through the first pupil region that is a split pupil corresponding to the subpixel S1, and a split pupil corresponding to the subpixel S2. The light beam ΦLb is incident on the sub-pixel S2 through the second pupil region. The light flux ΦLa is incident on the first photodiode 301A, and the light flux ΦLb is incident on the second photodiode 301B. Since these two light beams are incident from the same point on the subject, when the imaging optical system is in focus, the two light beams pass through the same microlens ML as shown in FIG. The matching sub-pixels S1 _and S2 are reached. Accordingly, the image signal of one column output from the column of the first sub-pixel S1 and the image signal of one column output from the column of the second sub-pixel S2 are substantially coincident with each other.

However, as shown in FIG. 7, in a state where the focus is shifted by x, the arrival positions of the two light beams ΦLa and ΦLb are shifted from each other by the change in the incident angle of the light beams ΦLa and ΦLb to the microlens. Therefore, light from the same point on the subject passes through different microlenses ML and reaches the subpixels S1 _and S2 separated on the image sensor, and is incident on the first and second photodiodes 301A and 301B that each has. To do. Therefore, a phase difference is generated between the image signal of one column output from the column of the first subpixel S1 and the image signal of one column output from the column of the second subpixel S2.

<Relationship between phase difference and defocus amount>
FIG. 8 is a diagram illustrating a relationship between the phase shift amount of the image signal and the defocus amount of the focal position of the lens group 101 in a pair of subpixel columns including the subpixel S1 column and the subpixel S2 column. As shown in FIG. 8, the defocus amount (vertical axis) monotonously increases as the phase shift amount (horizontal axis) of the image signal increases. Further, the direction of the phase shift corresponds to the direction of the defocus amount (closest side / infinite side).

  FIG. 9 is a schematic diagram showing an image signal when the defocus amount is small, and FIG. 10 is a schematic diagram showing an image signal when the defocus amount is large.

  Of the subject image in which the light beam divided on the pupil plane is secondarily formed on the pixel column, an image based on the image signal acquired in the pixel column of the subpixel S1 is acquired in the pixel column of the A pixel and the subpixel S2. An image based on the obtained image signal is referred to as a B image. The A image and the B image are formed with an offset with respect to the optical axis center position of the sub-pixels S1 and S2. Further, each offset of the A image and the B image is symmetric with respect to the optical axis center.

  FIG. 9 shows a case where the defocus amount is small, and FIG. 10 shows a case where the defocus amount is large. The offset with respect to the optical axis center position in FIG. 10 is larger than that in FIG. As described above, the phase difference between the two subject images formed on the column of the subpixel S1 and the column of the subpixel S2 changes according to the defocus amount of the focal position of the lens group 101. The phase difference can be converted into a defocus amount by a known method, and the distance to the subject can be calculated by a known method based on the relationship between the defocus amount and the magnification ratio of the lens group 101.

<Improving reliability by averaging the defocus amount>
Next, improvement in the reliability of the focus detection result by averaging the focus detection results by the phase difference detection method will be described.

  In focus detection by the phase difference detection method, the difference between the focus state of the camera and the focus position of the subject image is detected. For example, when a sufficient signal is not obtained with low luminance, the ratio of noise included in the signal of the subject image increases, so that the signal is distorted, and an error occurs in the correlation calculation when calculating the phase difference. As a result, the defocus amount varies and an error occurs. For example, when the signal level of the subject image is higher than the noise level, the signal of the subject image obtained by the focus detection pixel is not significantly deformed by noise, so the defocus amount based on the pair of parallax image signals is relatively accurate. Is required. In this case, by obtaining the defocus amount a plurality of times under the same condition and averaging the obtained results of the plurality of defocus amounts, variations in the defocus amount can be suppressed, and focus detection accuracy can be improved. On the other hand, when the noise level is higher than the signal level of the subject image, the subject image signal obtained by the focus detection pixel is significantly deformed by noise, and the defocus amount based on the pair of parallax image signals is irrelevant to the subject image. Often show incorrect results. In this case, even if a plurality of defocus amounts are acquired under the same conditions and the results are added and averaged, the average of the results not depending on the subject image is taken. Therefore, no matter how many result samples are added, the focus detection accuracy cannot be improved.

  Therefore, in the present embodiment, the focus detection accuracy is improved by replacing the correlation amount described later under the subject condition such as low luminance where the signal level of the subject image is lower than the noise level. It is also possible to use both the defocus amount addition average and the correlation amount replacement.

<Influence on the other image signal when noise is mixed in one image signal obtained by pupil division>
Next, the image signal and the correlation amount when noise is mixed in the first or second subpixel column will be described.

  FIG. 11 shows a pattern subject 1200 shown in a graphic pattern that is easy to understand for explanation. The pattern subject 1200 includes a gray portion 1201 and a white portion 1202. In FIG. 11, an area of a pixel column used for focus detection (so-called focus detection area), which is focused on for explanation, is 1203.

  FIG. 12 is a graph showing an A image signal (b), a B image signal (c), and an A + B image signal (a) for the pattern subject 1200 shown in FIG. It is assumed that the focus of the lens group 101 is in focus and no phase difference occurs between the A image signal and the B image signal.

FIG. 12A is a graph of an A + B image signal that is an addition signal, and the image signal output of the gray portion 1201 having a low luminance in the interval where the X coordinate in the horizontal direction is X _{0 to} X _{1 and} the interval X _{2 to} X _3. Is indicated by Y ₂ . Further, the image signal output of the white portion 1202 having a high luminance in the section from X _{1 to} X ₂ is indicated by Y ₄ . Since a constant signal value is added even in the absence of incident light quantity due to the influence of dark current generated in the imaging unit 213, the signal level is indicated by OB. FIG. 12B shows an image of the first sub-pixel column. FIG. 12C is a graph showing a B signal which is an image signal of the second sub-pixel column. In FIGS. 12B and 12C, the image signal output of the gray portion 1201 having a low luminance in the interval where the X coordinate in the horizontal direction is X _{0 to} X _{1 and} the interval X _{2 to} X ₃ is indicated by Y _1. Yes. Further, the image signal output of the white portion 1202 having a high luminance in the section from X _{1 to} X ₂ is indicated by Y ₃ . Here, first the image signal output Y ₁ of the second sub-pixel columns is an intermediate value of the summed image signal output Y ₂ and OB. The first and the image signal output Y ₃ of the second sub-pixel columns is an intermediate value of the summed image signal output Y ₄ and OB.

  FIG. 13 shows that independent noise is mixed in the image signal (b) of the first subpixel column and the addition signal (a) of the image signals of the first and second subpixel columns shown in FIG. It is the graph which showed the image signal (c) of the 2nd subpixel row | line in the case where it did.

  In the floating diffusion region 303, the imaging unit 213 according to the present embodiment performs voltage conversion on the charge of only the first subpixel column and voltage conversion on the charge of the addition signal of the first and second subpixel columns. Are different. For this reason, the noise added to both does not necessarily match. For this reason, FIG. 13 shows a case where independent noises are mixed.

13 (a) is a graph showing how the noise is mixed in the addition signal (A + B image signal), in the horizontal direction of the X coordinate of the X ₁ to X ₂ interval, is schematically noise .DELTA.N _AB is shown Yes. FIG. 13B is a graph showing a state in which noise is mixed in the image signal (A image signal) of the first sub-pixel row, and schematically shows a region where the horizontal X coordinate is X _{1 to} X _2. noise .DELTA.N _A is shown. FIG. 13 (c) is a graph showing the manner in which the noise on the image signal is mixed in the second sub-pixel columns, the horizontal direction of the X coordinate of the X ₁ to X ₂ interval, schematically noise .DELTA.N _B is shown Has been. Here, the image signal of the second subpixel column is obtained by electrically subtracting the image signal of the first subpixel column from the addition signal of the image signals of the first and second subpixel columns. For this reason, the image signal (B + ΔN _B ) of the second subpixel column is a signal represented by the following Expression 1.
B + ΔN _b = (A + B + ΔN _ab ) − (A + ΔN _a )
= B + ΔN _ab −ΔN _a
∴ΔN _b = ΔN _ab −ΔN _a (Expression 1)
A: A signal based on the charge generated by the first photodiode (a signal obtained by subtracting noise from the A image signal)
B: A signal based on the charge generated by the second photodiode (a signal obtained by subtracting noise from the B image signal)
ΔN _ab : Noise superimposed on the addition signal ΔN _a : Noise superimposed on the A image signal ΔN _b : Noise superimposed on the B image signal

Comparing (b) and (c) in FIG. 13, ΔN _a and ΔN _b are generated in the opposite direction, and the image signal is compared with the state of (b) and (c) in FIG. It can be seen that the correlation is worse.

FIG. 14 is a graph showing the correlation between the image signals (A image signal and B image signal) of the first and second subpixel columns in FIG. 12 for each phase shift amount. In FIG. 12, since the focus of the lens group 101 is in focus and there is no phase difference between the image signals of the first and second subpixel columns, two images are obtained when the phase shift amount is zero. they become a difference of the C ₁ minimum, highest correlation is high.

FIG. 15 shows the correlation between the image signals (A image signal and B image signal) of the first and second subpixel columns in FIG. 13 for each phase shift amount when noise is mixed in FIG. It is a graph. 15, as described with reference to FIG. 13 (b) and (c), respectively on the image signal noise .DELTA.N _B and .DELTA.N _a first and second sub-pixel columns are mixed. Due to the mixing of noise, the difference between the two images increases in the range of the value ± P with the phase shift amount sandwiching zero. Although the difference between the two images is the amount of phase shift was C ₂ in value ± P, the difference between the two images in the phase shift amount zero due to noise .DELTA.N _B and .DELTA.N _a is increased to [Delta] C _N worsening would value C ₃ It has been. If such a correlation amount is obtained, the phase shift amount -P or + P as shown in FIG. 15 is most likely to be judged as having the highest correlation when the phase shift amount is zero as shown in FIG. An error occurs when it is determined that the correlation is high. If an error occurs in the correlation amount in this way, the accuracy of focus detection may be deteriorated.

  Therefore, in this embodiment, the correlation amount in a part of the acquired correlation calculation result is replaced by the correlation amount interpolation unit 230, and the phase difference at which the correlation amount has the minimum value is acquired using the correlation amount after replacement. And get the defocus amount. Details of the correlation amount interpolation (replacement of correlation amount) by the correlation amount interpolation unit 230 will be described together with the following description of the operation of the digital camera system.

<Operation of digital camera system>
FIG. 16 is a flowchart showing a flow of focus detection operation of the digital camera system in the present embodiment. This operation is realized by the camera controller 215 controlling each unit.

  In the digital camera system according to the present embodiment, the operation mode is automatically set to the live view mode after the power is turned on, and the image signal (A image signal) of the first subpixel column is continuously captured by the imaging unit 213. ) And two image addition signals (A + B image signals) are generated. In addition, it is assumed that the captured subject image is displayed on the display unit 224 using the addition signal.

  First, the focus detection operation is started when SW1 of the operation switch 214 is pressed and a focus detection command is generated.

  In step S <b> 1701, the camera controller 215 transfers the image signal of the first subpixel column and the addition signal of the image signals of the first and second subpixel columns from the imaging unit 213 to the pixel subtraction unit 228. (Hereinafter, the image signals of the first and second subpixel columns are referred to as two images.) The pixel subtracting unit 228 electrically subtracts the image signal of the first subpixel column from the addition signal of the two images. As a result, the image signal (B image signal) of the second sub-pixel row is calculated. Thereby, independent image signals of the first and second subpixel columns are obtained. After acquisition, the process proceeds to step S1702.

  In step S <b> 1702, the camera controller 215 performs correction processing for suppressing various signal level fluctuations such as black level correction and a drop in peripheral light amount caused by the imaging optical system, independently for each of the two images. The reason why the signal level fluctuates is that light rays having different angles are incident on the first subpixel row and the second subpixel row by dividing the subject image into light rays on the exit pupil plane. After the correction process, the process proceeds to step S1703.

  In step S <b> 1703, the camera controller 215 transfers the two images from the pixel subtraction unit 228 to the correlation calculation unit 219. The correlation calculation unit 219 sets each shift amount in the first phase shift range including zero phase shift amount (also referred to as a position where the pair of parallax image signals are not shifted) set based on the defocus range in which the focus is detected. The difference between the two images is calculated as a correlation amount. This correlation amount indicates the degree of mismatch between the two images, and the lower the mismatch degree, the higher the correlation between the two images. After the calculation, the process proceeds to step S1704.

  Steps S1704 to S1710 are characteristic operations of the present embodiment in which a part of the correlation amount is interpolated from the surrounding correlation amounts as necessary with respect to the calculated correlation amount, and will be described in detail below.

  In step S1704, the camera controller 215 transfers the two images to the correlation amount interpolation unit 230. The correlation amount interpolation unit 230 calculates the height difference (so-called contrast) from the highest value and the lowest value of the pixel signals included in the two images. As the difference in height decreases, it becomes more likely that the change in the correlation amount due to the noise cannot be ignored with respect to the change in the correlation amount due to the waveform of the two images. For this reason, the height difference is calculated in this step. After the calculation, the process proceeds to step S1705.

  In step S1705, the correlation amount interpolation unit 230 determines whether or not the height difference calculated in step S1704 is less than a predetermined threshold value. If it is less than the threshold value, the process proceeds to step S1706 to perform the interpolation processing described in the subsequent step. If it is equal to or greater than the threshold value, it is determined that the interpolation process may be omitted, and the process proceeds to step S1711.

  In step S1706, the correlation amount interpolation unit 230 determines a second phase shift range that is a range for replacing the correlation amount. A method for determining the second phase shift range in this step will be described. First, the correlation amount interpolation unit 230 performs an operation of searching for the phase shift amount ± P in FIG. In other words, in both positive and negative directions with the phase shift amount zero interposed therebetween, a phase shift amount in which the absolute value is as small as possible and the correlation amount takes a minimum value is searched. Then, the correlation amount interpolation unit 230 sets the phase shift range (a range larger than −P and smaller than + P in FIG. 15) sandwiched between the detected positive and negative phase shift amounts to the phase shift range (second phase) to be interpolated. Shift range). After determination, the process proceeds to step S1707.

  In step S1707, the correlation amount interpolating unit 230 searches for a phase shift condition that minimizes the correlation amount, extracts a condition that minimizes the correlation amount, and differentiates the correlation amount under the phase shift condition. Is calculated as the correlation steepness. After the calculation, the process proceeds to step S1708.

  In step S1708, the correlation amount interpolation unit 230 determines whether the correlation steepness calculated in step S1707 is equal to or greater than a predetermined threshold. If it is equal to or greater than the threshold value, it can be seen that the correlation amount changes sharply with a slight phase shift fluctuation. In this case, when interpolating the correlation amount in the subsequent step, if the interpolation is performed with the correlation amount of the shift amount adjacent to the range of the phase shift amount to be interpolated, a sharp change in the correlation amount cannot be reproduced. . Thus, an error occurs in the determination of the phase shift amount with the highest correlation, and it is difficult to improve the accuracy of focus detection performed using the interpolated correlation amount. In order to avoid such a problem, interpolation based on a sharp change in correlation is performed in the subsequent stage (S1709). Therefore, if it is equal to or greater than the threshold value, the process advances to step S1709 for interpolation based on a steep change in correlation with a heavy calculation load. If the correlation steepness is less than the threshold, the process advances to step S1710 where simple interpolation is performed with a light calculation load.

  In step S1709, a mathematical expression representing the relationship between the correlation amount and the phase shift amount is acquired based on the relationship between the correlation amount and the phase shift amount in the shift range where the correlation amount is not replaced. Then, the correlation amount in the second phase shift range acquired using this mathematical formula is set as the correlation amount in the second phase shift range (first interpolation method). Specific processing will be described. The correlation amount interpolation unit 230 sets the range of the predetermined shift amount in the direction opposite to the phase shift amount zero from the phase shift amount (± P) searched in step S1706 as the starting phase shift range (first shift). 3 may be referred to as a phase shift range of 3). Then, the correlation amount interpolation unit 230 uses a known interpolation method such as spline interpolation or least square method using the relationship between the phase shift amount of the third phase shift range and the correlation amount, and uses a mathematical expression (interpolation). (Sometimes called an expression). The mathematical formula is preferably a higher order formula. The width of the third phase shift range is appropriately set according to the slope of the graph of the correlation amount and the phase shift amount, the magnitude of the image signal (S / N ratio), the accuracy of the interpolation formula to be acquired, the balance of calculation addition, etc. can do. For example, the third phase shift range may be a range from the phase shift amount (± P) searched in step S1706 to a range where the slope of the graph takes a predetermined value or more. After the interpolation formula is acquired, the interpolation process is performed by applying the acquired interpolation formula to the second phase shift range that is the phase shift amount range of the interpolation destination, and the correlation amount in the second phase shift range is acquired. . Then, the correlation calculation unit 219 acquires the correlation amount in the second phase shift range by replacing the correlation amount acquired in step S1703 with the acquisition result acquired in step S1709 (correlation amount acquired by the interpolation process). Replace part of the correlation calculation curve. After the replacement, the process proceeds to step S1711.

  In step S1710, simple interpolation is performed with a lighter calculation load than in step S1709, and the correlation amount in the second phase shift range is replaced with the minimum value of the correlation amount (second interpolation method). Specific processing will be described. The correlation amount interpolating unit 230 uses the second phase shift range determined in step S1706 as the phase shift amount correlation amount (correlation amount at + P and correlation amount at −P) serving as the boundary of the second phase shift range. Replace with the smaller correlation amount. Thereby, a part of the correlation calculation curve acquired by the correlation calculation unit 219 is replaced.

  The phase shift amount serving as the boundary of the second phase shift range refers to the phase shift range serving as the boundary between the range where replacement is performed and the range where replacement is not performed, and the correlation amount searched in step S1706 takes a minimum value. It matches the phase shift amount. After the replacement, the process proceeds to step S1711.

  FIG. 17A is a graph of the correlation amount and the phase shift amount obtained by performing the first interpolation method in step S1709. FIG. 17B is a graph of the correlation amount and the phase shift amount obtained by performing the second interpolation method in step S1710. In each graph, the dotted line indicates the relationship between the correlation amount before interpolation and the phase shift amount, and the solid line indicates the relationship between the correlation amount after interpolation and the phase shift amount. As shown in FIGS. 17 (a) and 17 (b), before the interpolation, the influence of noise appears in the range of ± P as in the graph shown in FIG. 15, but the first or second interpolation method is performed. As a result, it is possible to reduce the influence of noise appearing in the range of ± P.

  In step S1711, the camera controller 215 calculates the maximum correlation shift amount with the smallest correlation amount based on the correlation calculation curve. The correlation calculation curve used by the camera controller 215 under the conditions for performing the interpolation in step S1709 or step S1710 is a correlation calculation curve partially replaced by the correlation amount interpolation unit 230. On the other hand, under the condition of proceeding directly from step S1705 to step S1711, the correlation calculation curve used by the camera controller 215 is the correlation calculation curve acquired by the correlation calculation unit 219. Since the maximum correlation shift amount is rough as the accuracy used for focus detection in the unit of one shift amount, which is a natural number, it is preferable to obtain at least the third decimal point and preferably the fifth decimal point from the above interpolation formula. When the second interpolation method is used, the correlation amount becomes equal in the second phase shift range where the correlation amount is replaced. When the correlation amount in the second phase shift range takes the minimum value of the correlation amount in the first phase shift range, which is a range for calculating the correlation amount, the median value of the second phase shift range is set as the maximum correlation shift amount. can do.

  After calculating the maximum correlation shift amount, the process proceeds to step S1712.

  In step S1712, the camera controller 215 calculates a defocus amount based on the most correlated shift amount calculated in step S1711 and the optical characteristics of the lens unit 100. After calculation, the process proceeds to step S1713.

  In step S1713, the camera controller 215 controls the focal position of the lens group 101 via the lens controller 105 based on the calculated defocus amount. When the control is completed, the focus detection operation ends.

  According to this embodiment, the image signal of the second subpixel column is electrically acquired from the image signal of the first subpixel column and the addition signal of the image signals of the first and second subpixel columns. Therefore, the focus detection can be performed with high accuracy by reducing the influence of noise.

(Modification)
Here, a modified example of steps S1704 to S1705 will be described.

  In step S1704, the correlation amount interpolation unit 230 calculates the highest value and the lowest value in the two images, and performs the condition determination in step S1705. However, the present invention is not limited to this, and the condition determination in step S1705 may be performed according to various imaging conditions. For example, information regarding the magnitude (S / N ratio) of a signal with respect to noise included in an image signal obtained by imaging can be used. As information on the magnitude of the signal, the correlation amount interpolation unit 230 calculates the maximum value of the secondary differential value that is the maximum value of the result of second-order differentiation of the signal waveform of the two images, or the adjacent difference value of the signal waveform, It can be used for condition determination in step S1705. The adjacent difference value of the signal waveform refers to a value (local differential value) obtained by taking the difference between each of the plurality of signals constituting the signal waveform and the adjacent signals. For example, when the maximum value of the second derivative (sometimes referred to as the second derivative maximum value) or the adjacent difference value is less than a predetermined threshold, the correlation due to noise is correlated with the change in the correlation amount due to the waveform of the two images. Changes in quantity may not be negligible. In this case, since the accuracy of the interpolation formula acquired from the third phase shift range due to the influence of noise may be lowered, the second interpolation method described in step S1710 is more than the first interpolation method described in step S1709. It is preferable to perform an interpolation method.

  Further, the present invention is not limited to this, and various settings of the imaging means may be used as various imaging conditions. For example, the camera controller 215 outputs the ISO sensitivity setting, which is the exposure setting of the digital camera, and the exposure correction amount to the correlation amount interpolation unit 230, and the correlation amount interpolation unit 230 determines the condition in step S1705 based on these values. You may go. For example, when the ISO sensitivity setting is a predetermined threshold value or more, or when the exposure correction amount is set to be lower than the appropriate exposure, the waveform of the two images becomes smaller due to underexposure, and the correlation amount changes due to the waveform of the two images. On the other hand, there is a possibility that a change in correlation amount due to noise cannot be ignored. Also in this case, since the accuracy of the interpolation formula may be lowered due to the influence of noise, it is preferable to perform the second interpolation method rather than the first interpolation method.

  In addition, the temperature of the imaging unit may be used as various imaging conditions. For example, a temperature measurement unit that measures the temperature of the imaging unit 213 outputs temperature information that is the measurement result to the correlation amount interpolation unit 230, and the correlation amount interpolation unit 230 performs the condition determination in step S1705 based on the temperature. Also good. Since thermal noise increases as the temperature of the imaging unit 213 rises, noise is likely to be mixed into the image signal, and the change in the correlation amount due to the noise cannot be ignored with respect to the change in the correlation amount due to the waveform of the two images. It is because there is sex. In this case, since the accuracy of the interpolation formula may be lowered due to the influence of noise, it is preferable to perform the second interpolation method rather than the first interpolation method.

  In addition, the size of a predetermined spatial frequency component included in an image signal obtained by imaging may be used as various imaging conditions. For example, vertical stripe noise may occur in the imaging unit 213 due to current fluctuations generated as high-frequency noise from each electrical component of the digital camera. With this in mind, a BPF unit (not shown) that extracts a specific spatial frequency component is provided. If the output of the BPF part is equal to or greater than a predetermined threshold value, vertical stripe noise is mixed in the two images with a phase difference of zero, and there is a possibility that it cannot be ignored in the same manner as described above. Also in this case, since the accuracy of the interpolation formula may be lowered due to the influence of noise, it is preferable to perform the second interpolation method rather than the first interpolation method.

  Here, a modified example of step S1702 will be described.

  In step S12702, the camera controller 215 performs correction processing for suppressing various signal level fluctuations, such as black level correction and a decrease in peripheral light amount caused by the imaging optical system, independently for each of the two images. However, the present invention is not limited to this, and in addition to the above correction processing, the focus detection area is expanded in the vertical direction, and the pixel adder 217 generates a plurality of image signals of the first and second focus detection pixel arrays for each horizontal coordinate. Noise may be reduced by adding lines.

  In addition to this, in addition to the above correction processing, the LPF processing unit 229 may perform LPF processing on two images in the horizontal direction to reduce noise. However, if the frequency is too low, the edge component of the subject image may be dulled, which may reduce the accuracy of the correlation calculation. For example, a light LPF such as an adjacent average is preferable.

  The present invention is not limited to this, and the above-described LPF processing may be performed only when it is determined in step S1705 that the process proceeds to step S1706.

  By performing the operation as described above, it is possible to appropriately determine whether or not to perform the noise countermeasure operation.

  Here, a modified example of step S1703 will be described.

  In step S1703, the correlation calculation unit 219 calculates the difference between the two images for each shift amount as a correlation amount within a predetermined phase shift amount range including zero phase shift amount based on the focus detection defocus range. It was. However, the present invention is not limited to this, and in addition to the correction processing described above, the focus detection area is expanded in the vertical direction, and the correlation amount adding unit 220 calculates the correlation amounts for a plurality of lines calculated by the correlation calculation unit 219 as phase shift amounts. It is also possible to reduce the variation in the correlation amount by adding each time.

  Further, the present invention is not limited to this, and the correlation calculation unit 219 may calculate or output only the correlation amount of the odd phase shift amount. By not outputting even (including zero) phase shift amounts, it is possible to simply prevent the output of a zero phase shift correlation amount that is strongly influenced by noise. In this case, before step S1703, it is determined whether or not the condition for noise countermeasure is satisfied. If the condition is for noise countermeasure, it is preferable to calculate or output only the correlation amount of the odd phase shift amount. The correlation amount in even-numbered phase shift amounts including zero phase shift amount can be interpolated using the correlation amount in odd-numbered phase shift amounts using the method of step S1709 or step S1710. In this case, since the correlation amount at the even phase shift amount is not output before the interpolation, the correlation amount represented by the dotted line shown in FIG. 17 is not replaced with the correlation amount represented by the solid line. However, since the defocus amount is acquired using the correlation amount acquired by interpolation instead of the correlation amount acquired by correlation calculation, it is considered that the correlation amount acquired by correlation calculation is replaced with the correlation amount acquired by interpolation. Shall. As described above, when the correlation calculation unit 219 outputs only the correlation amount of the odd number of phase shift amounts, the second phase shift range is a range between ± 1, but other ranges (for example, if necessary) (for example, The range between +1 and +3) may also be interpolated as a replacement target.

  In the above-described embodiment, the correlation amount interpolation unit 230 can execute a plurality of interpolation methods, and selects an interpolation method to be executed according to the relationship between the correlation amount and the phase shift amount. The unit may execute only one interpolation method. In addition, in the above-described embodiment, the case where focus detection is performed by dividing the pixels of the image sensor into two (sub-pixels S1 and S2) has been described, but the number of divided pixels is not particularly limited.

  In the present embodiment, all or a part of the functions shown in FIG. 16 may be realized by dedicated hardware, or a program for realizing the functions of each process may be recorded on a storage medium. The program may be read by the focus detection device and executed. The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

  A program product such as a computer-readable recording medium in which the above program is recorded can also be applied as an embodiment of the present invention. The above program, recording medium, transmission medium, and program product are included in the scope of the present invention.

  As mentioned above, although preferable embodiment of this invention was described, this invention is not limited to these embodiment, A various deformation | transformation and change are possible within the range of the summary.

  As mentioned above, although preferable embodiment of this invention was described, this invention is not limited to these embodiment, A various deformation | transformation and change are possible within the range of the summary.

DESCRIPTION OF SYMBOLS 100 Lens part 101 Lens group 102 Diaphragm | restriction 108 Electric contact unit 200 Camera part 213 Imaging part 214 Operation switch 224 Display part 230 Correlation amount interpolation part ML Micro lens S1 1st focus detection pixel S2 2nd focus detection pixel 501 Smooth layer 502 Light shielding layer PRES Reset pulse signal PTXA First transfer pulse signal PTXB Second transfer pulse signal PSEL Vertical selection pulse signal 1200 Pattern subject 1201 Gray portion of pattern subject 1202 White portion of pattern subject 1203 Focus detection area 301A First photodiode 301B Second photodiode 302A First transfer switch 302B Second transfer switch 303 Floating diffusion region 304 Amplifying unit 305 Reset switch 306 Selection Switch

Claims (11)

A first acquisition means for performing a correlation calculation while shifting a pair of parallax image signals within a first phase shift range and acquiring a correlation calculation curve;
Among the correlation calculation curves of the first phase shift range acquired by the first acquisition means, the correlation between the pair of parallax image signals is within a range sandwiching positions where the pair of parallax image signals are not shifted. Replacement means for replacing the correlation calculation curve in the second phase shift range, which is a range sandwiched between minimum values to be increased;
And a second acquisition unit that acquires a defocus amount from the correlation calculation curve partially replaced by the replacement unit. The replacement means includes
Replacing the correlation amount in the second phase shift range by setting the value of the correlation amount in the phase shift amount serving as the boundary of the second phase shift range as the correlation amount in the second phase shift range. The focus detection apparatus according to claim 1, wherein The replacement means includes
Of the first phase shift range, the second phase is calculated using an equation obtained based on the relationship between the correlation amount and the phase shift amount in a third phase shift range that is not the second phase shift range. Get the correlation amount in the shift range,
The focus detection apparatus according to claim 1, wherein a correlation amount in the second phase shift range is replaced using an acquisition result. The replacement means includes
4. The method of replacing a correlation amount in the second phase shift range is selected based on a relationship between the correlation amount and the phase shift amount in the first phase shift range. The focus detection apparatus according to claim 1. According to the imaging conditions under which the pair of parallax image signals are acquired,
Control means for determining whether or not to perform the replacement by the replacement means;
When the control unit determines that the replacement unit does not perform the replacement, the second acquisition unit acquires a defocus amount using a correlation calculation curve that has not been replaced. Item 5. The focus detection apparatus according to any one of Items 1 to 4.   6. The focus detection according to claim 1, wherein the replacement unit determines the second phase shift range in accordance with an imaging condition from which the pair of parallax image signals are acquired. apparatus.   The imaging conditions include ISO sensitivity in the imaging means, a maximum value of signals included in the pair of parallax image signals, a difference between a maximum value and a minimum value of signals included in the pair of parallax image signals, and the pair of parallaxes By the maximum value of the second derivative value of the signal included in the image signal, the maximum value of the adjacent difference value of the signal included in the pair of parallax image signals, the exposure setting in the imaging unit, and the temperature measurement unit that measures the temperature of the imaging unit The focus detection apparatus according to claim 5, wherein the focus detection apparatus is at least one of a measurement result and a magnitude of a predetermined spatial frequency component included in the pair of parallax image signals. A first photoelectric converter that receives a light beam that passes through a first pupil region of an exit pupil of the imaging optical system; and a second that receives a light beam that passes through a second pupil region of the exit pupil of the imaging optical system. An imaging means having a plurality of pixels each having a photoelectric conversion unit;
Image processing for generating the pair of parallax image signals from first image signals obtained from the plurality of first photoelectric conversion units and second image signals obtained from the plurality of second photoelectric conversion units. The focus detection apparatus according to claim 1, further comprising: an apparatus. The imaging means includes a first image signal obtained from the plurality of first photoelectric conversion units, an addition signal obtained from the plurality of first photoelectric conversion units and the plurality of second photoelectric conversion units, and , Output
The image processing means acquires a second image signal obtained from a plurality of the second photoelectric conversion units by taking a difference between the addition signal and the first image signal. Item 9. The focus detection apparatus according to Item 8. Obtaining a correlation calculation curve by performing a correlation calculation while shifting a pair of parallax image signals within a first phase shift range;
Of the correlation calculation curve, the second phase shift range of the second phase shift range located between a pair of local minimum values where the correlation between the pair of parallax image signals is high and is located across the position where the pair of parallax image signals is not shifted. Replacing the correlation calculation curve; and
And a step of acquiring a defocus amount from the correlation calculation curve partially replaced.   A computer-readable program for causing a focus detection apparatus to execute the focus detection method according to claim 10.

Images