JP5950679B2 - FOCUS DETECTION DEVICE, IMAGING DEVICE, AND FOCUS DETECTION METHOD - Google Patents

Description

  The present invention relates to a technique for performing phase difference focus detection using signals of pupil-divided images acquired from a plurality of photoelectric conversion units sharing a microlens. In particular, the present invention relates to a technique for avoiding an influence caused by a saturation state of a signal of a pupil division image when a correlation calculation (phase difference detection calculation) is performed by applying a band pass filter to the signal of the pupil division image.

  For an imaging apparatus such as a digital still camera, a photoelectric conversion element (PD: photodiode) arranged so as to constitute a unit pixel with respect to one microlens constituting a solid-state imaging element is divided into a plurality of parts. At the same time, a technique for performing focus detection by a phase difference method is known.

  For example, the photoelectric conversion elements constituting the unit pixel are divided into two (hereinafter, the divided individual photoelectric conversion elements are referred to as “divided PDs”, and the pixels corresponding to the divided PDs are referred to as “divided pixels”). A technique has been proposed in which the charge of one divided PD is read nondestructively, then the added charge of the two divided PDs is read, and the charge of the other divided PD is obtained from the difference between the two charges thus read ( Patent Document 1). That is, after reading the pixel value of one divided pixel, the addition value of two divided pixels (pixel value of a unit pixel) is read, and the pixel value of the other divided pixel is obtained from the difference between them. As a result, while maintaining the high sensitivity characteristics of the sum signal of the output signals from the two divided PDs used as the image signals for imaging, the output signals from the two divided PDs used as the phase difference image signals (the pupil divided image). Signal).

  Further, by removing the DC component of the output signals (pupil divided image signals) from the plurality of divided PDs by a band-pass filter (hereinafter referred to as “BPF”), it is possible to detect the focus well even when a level difference due to ghost occurs. A possible technique has been proposed (see Patent Document 2).

Japanese Patent No. 469930 Japanese Patent No. 2762463

  However, in the case of the technique described in Patent Document 1, the linearity is lost when the output signals added from the plurality of divided PDs exceed the range of A / D conversion, and have hard knee characteristics. When the technique described in Patent Document 2 is applied to this characteristic, the distortion corresponding to the number of taps of the BPF remains in the divided PD adjacent to the saturated divided PD, so that the reliability of the focus detection calculation is lowered. Resulting in.

  An object of the present invention is to provide a technique for avoiding the influence of saturation of divided PDs when a correlation calculation is performed by filtering a phase difference image signal for focus detection with a BPF.

The focus detection apparatus according to the present invention adds an image sensor having a plurality of photoelectric conversion units sharing a microlens, an output signal from the plurality of photoelectric conversion units, and an output signal from the plurality of photoelectric conversion units. Generating means for generating an addition signal, detection means for detecting a saturated pixel region of the image sensor from the addition signal, frequency removing means for removing a predetermined frequency from output signals from the plurality of photoelectric conversion units, A saturation pixel region detected by the detection means, a detection signal generation means for generating a saturation detection signal in accordance with the processing by the frequency removal means, and a signal or mask signal processed by the frequency removal means based on the saturation detection signal And a defocus amount calculator that calculates a defocus amount based on the phase difference image signal output from the selection unit. And means, wherein the detection signal generating means generates a first saturation detection signal indicating the saturation pixel area detected by the detection unit, the range shown a saturated pixel area in the first saturation detection signal When the output signals from the plurality of photoelectric conversion units are processed by the frequency removing unit, a second saturation detection signal expanded to a range in which the influence of distortion is generated is generated, and the selecting unit is used as the phase difference image signal. Switching is performed such that the mask signal is output at the rising edge of the second saturation detection signal, and the signal processed by the frequency removing unit is output at the falling edge of the second saturation detection signal. To do.

  According to the present invention, when the phase detection image signal for focus detection is filtered by the BPF and correlation calculation is performed, the influence of saturation in the divided PD can be avoided, thereby performing accurate focus detection. be able to.

1 is a block diagram illustrating a schematic configuration of an imaging apparatus according to an embodiment of the present invention. It is the front view and sectional drawing of a unit pixel cell which comprise the image sensor of FIG. It is a block diagram which shows the detailed structure of the AF evaluation value calculation part shown in FIG. It is a graph which shows the output characteristic with respect to the exposure amount of division | segmentation PD shown in FIG. It is a figure which shows typically the influence by the pixel level by the presence or absence of saturation in the division | segmentation PD shown in FIG. 2, and a band pass filter with a signal waveform. FIG. 4 is a timing chart of various signals when the filter tap length of the bandpass filter shown in FIG. 3 is 5 taps. FIG. 4 is a circuit diagram of the bandpass filter shown in FIG. 3. It is a graph which shows the filter characteristic (relationship between a spatial frequency and the transmittance | permeability) of the band pass filter shown in FIG. It is a figure which shows the output signal waveform of the selection selector part shown in FIG. 3 is a timing chart schematically showing a process of calculating a B signal in the B image generation unit shown in FIG. 1. It is a timing chart which shows the timing which the NG determination part shown in FIG. 3 outputs a NG determination signal. It is a block diagram of the AF evaluation value calculation part which concerns on another embodiment of this invention.

<Overall configuration of imaging device>
FIG. 1 is a block diagram illustrating a schematic configuration of an imaging apparatus 100 according to an embodiment of the present invention. In the imaging apparatus 100, an image in real space is formed on an imaging element 102 that converts an optical image into an electrical signal through an optical system including an imaging lens group 101 including a focusing lens. The image sensor 102 has a structure in which a plurality of divided PDs (a plurality of photoelectric conversion units constituting a unit pixel cell) sharing a microlens are arranged in a Bayer array, and each divided PD performs charge accumulation, The data is output to the A / D conversion unit 103.

  A unit pixel cell of the image sensor 102 will be described with reference to FIG. FIG. 2A is a front view of a unit pixel cell constituting the image sensor 102, and FIG. 2B is a cross-sectional view of the unit pixel cell of FIG. As shown in FIG. 2A, a PD (photodiode) 201 that is a unit pixel cell includes a microlens 204, a divided PD 202 and a divided PD 203 corresponding to two divided pixels. Share the microlens 204. Further, the PD 201 further includes a color filter 205 and a wiring layer 206 as shown in FIG.

  Incident light passes through the microlens 204, has spectral characteristics by the color filter 205, and is irradiated to the divided PDs 202 and 203. Each of the divided PDs 202 and 203 has a characteristic in which the light flux of the exit pupil is limited. The divided PD for A image (hereinafter referred to as “A divided PD 202”) and the divided PD for B image (hereinafter referred to as “B image used”). Used as “divided PD 203”).

  Signals output from the A-image divided PD 202 and the B-image divided PD 203 are subjected to predetermined processing described later, and then are described later as focus detection phase difference image signals (hereinafter simply referred to as “phase difference image signals”). To the phase difference detection unit 107. The phase difference detection unit 107 performs focus detection processing using a phase difference detection method.

  The electric charge accumulated in each PD 201 of the image sensor 102 is read out by a control unit (not shown). The analog signal output from the image sensor 102 in this manner is input to the A / D conversion unit 103 via a CDS circuit (not shown) or an analog signal processing unit typified by ISO sensitivity amplification. The A / D conversion unit 103 converts the input analog signal into an A / D conversion unit output signal 116 that is a digital signal, and outputs the signal to the separation unit 112.

  The separation unit 112 generates an A signal 118 as a phase difference image signal and an A + B signal 117 as an imaging image signal (addition signal) from the A / D conversion unit output signal 116 received from the A / D conversion unit 103. And selectively output these signals. Specifically, the separation unit 112 outputs the A signal 118 to the delay line unit 114 of the B image generation unit 113, and the A + B signal 117 as the digital signal processing unit 104 and the detection signal generation unit of the AF evaluation value calculation unit 300. 109 and the B image calculation unit 115 of the B image generation unit 113. The A signal 118 is a signal corresponding to the output signal from the A image division PD 202, and the A + B signal 117 is a signal obtained by adding the output signals from the A image division PD 202 and the B image division PD 203. Signal.

  The B image generation unit 113 includes a delay line unit 114 and a B image calculation unit 115. The B image generation unit 113 subtracts the A signal 118 from the A + B signal 117 to generate a B signal 120 that is a phase difference image signal. Bear. That is, the B signal 120 is a signal corresponding to the output signal from the B image division PD 203. Specifically, the delay line unit 114 delays the A signal 118 by one line to generate a delayed A signal 119, outputs it to the B image calculation unit 115 in synchronization with the timing of the A + B signal 117, and performs AF evaluation. The data is output to the filter unit 111 of the value calculation unit 300. The B image calculation unit 115 subtracts the delay A signal 119 from the A + B signal 117 to generate a B signal 120, and outputs the B signal 120 to the filter unit 111 of the AF evaluation value calculation unit 300.

  A process for calculating the B signal 120 by subtracting the delay A signal 119 from the A + B signal 117 will be described with reference to the timing chart of FIG. In the operation of the first line, the A / D converter output signal 116 output from the A / D converter 103 is separated into the A + B signal 117 and the A signal 118 by the separator 112.

  In order to match the timing of the A + B signal 117 and the A signal 118 to the B image calculation unit 115, the A signal 118 is output to the B image calculation unit 115 as the delayed A signal 119 via the delay line unit 114. . The B signal 120 is obtained by subtracting the delayed A signal 119 from the A + B signal 117 by the B image calculation unit 115. Such an operation is similarly repeated from the second line onward.

  Returning to the description of FIG. The AF evaluation value calculation unit 300 is a characteristic block that constitutes a focus detection device in the imaging device 100. The AF evaluation value calculation unit 300 includes a filter unit 111, a detection signal generation unit 109, an NG determination unit 110, and a phase difference detection unit 107. The AF evaluation value calculation unit 300 generally calculates a defocus amount that is an AF evaluation value based on an input signal, and generates an NG determination signal that determines whether or not to perform an AF operation. The amount and the NG determination signal are output to the AF control unit 106. A more specific description will be given later.

  The AF control unit 106 determines whether to move the focusing lens to the in-focus position based on the defocus amount and the NG determination signal output from the AF evaluation value calculation unit 300, and performs drive control of the lens group 101. . The digital signal processing unit 104 performs digital signal processing such as image interpolation processing and development processing on the output of the separation unit 112 and outputs the result to an external storage device via a DRAM (not shown).

<Configuration of AF Evaluation Value Calculation Unit 300>
FIG. 3 is a block diagram illustrating a detailed configuration of the AF evaluation value calculation unit 300. As shown in FIG. 1, the AF evaluation value calculation unit 300 is generally configured by a filter unit 111, a detection signal generation unit 109, an NG determination unit 110, and a phase difference detection unit 107. The AF evaluation value calculation unit 300 generates and outputs a defocus amount and an NG determination signal from the A + B signal 117 that is an image signal for imaging and the delayed A signal 119 and the B signal 120 that are phase difference image signals.

[Configuration of Detection Signal Generation Unit 109]
In order to facilitate understanding of the function of the detection signal generation unit 109, the influence of saturation of the divided PD will be described with reference to FIG. 4 before describing the function of the detection signal generation unit 109. FIG. 4 is a graph showing the output characteristics of the A image division PD 202 and the B image division PD 203 with respect to the exposure amount and the output characteristics of the A + B image. FIG. 4A shows the ISO 100 case, and FIG. ) Shows the case of ISO200.

  In the case of FIG. 4A, when exposure is started, the output values from the A image division PD 202 and the B image division PD 203 increase linearly as the exposure amount increases for the A image, the B image, and the A + B image. To do. For the A image, the output value approaches the saturation level near the point P. Then, the A image division PD 202 leaks electric charge to the B image division PD 203 shared by the microlenses 204. Due to this influence, the linearity shown by the broken line is broken for the A image, and the output from the A image division PD 202 gradually approaches saturation, while the linearity shown by the broken line is broken for the B image, The output of the B image division PD 203 increases nonlinearly. Thus, the output characteristics for each of the A and B images (respective output characteristics of the A image division PD 202 and the B image division PD 203) exhibit soft knee characteristics in the vicinity of the point P.

  Further, if the exposure is further continued beyond the point P, the respective outputs for the A + B image and the A image are saturated at the point Q. Therefore, the B image obtained by subtracting the output for the A image from the output for the A + B image. Similarly, the output for saturates at the Q point. Thus, the output for each of the A image, the B image, and the A + B image shows a hard knee characteristic that is a nonlinear characteristic on the graph.

  In the case of FIG. 4B, the output gain is increased by an analog signal processing unit (not shown) in order to increase the exposure sensitivity of the A-image divided PD 202, and accordingly, the exposure sensitivity of the A + B image is also increased. It is done. When the exposure is started, the A image is output at a point H exceeding the saturation level of the A image in ISO 100. When the output for the A + B image reaches a saturation level (equivalent to the A + B image saturation level in FIG. 4A) at the point I, which is the upper limit of the conversion range of the A / D conversion unit 103, the output from the A + B image is converted to the A image. The output for the B image minus the output for is reduced. Therefore, the output for the B image has nonlinear characteristics at the point I, and gradually decreases from the point I so as to approach 0 at the point R.

  As described above, the output characteristics for the A image and the B image (the output characteristics of the A image division PD 202 and the B image division PD 203) are nonlinear. Therefore, the BPF (band pass) is converted into pixel values by A / D conversion. When the filter process is performed with a filter, distortion occurs. Distortion by the filter processing in the BPF will be described with reference to FIG.

  FIG. 5 is a diagram schematically showing the signal level and the influence of the pixel level and the BPF depending on the presence or absence of saturation in the A image division PD 202 and the B image division PD 203. In FIG. 5A, the horizontal scanning line position of the image sensor 102 is taken on the horizontal axis, and the pixel value level is taken on the vertical axis, and the imaging image signal (A + B image) and the phase difference image signal (A image, B image). ) Saturation characteristics. It is assumed that the A image, the B image, and the A + B image are saturated near the center of one line.

  When scanning is started, the pixel values of the A, B, and A + B images gradually approach the saturation level. Above the pixel value at the point E, the A image has a value lower than the ideal pixel value because the leakage of the charge amount described above occurs. When the pixel value of the A + B image reaches the saturation level, the pixel values of the A image, the B image, and the A + B image are saturated at the point V and stick to the upper limit value (saturation level) in the A / D conversion. As a result, the A image and the B image exhibit hard knee characteristics, and distortion is generated by the filter processing by the BPF.

  Similarly, when the pixel value of the A + B image goes out of the saturation level at the point W, the A image becomes a value lower than the ideal pixel value due to the influence of charge leakage. On the other hand, the B image has a higher value than the ideal pixel value. Therefore, both the A image and the B image have a hard knee characteristic when the pixel value of the A + B image goes out of the saturation level.

  FIG. 5B shows the charge leakage level of the A image and the B image between the point V and the point W (saturation range), which is the saturation region of the pixel value of the A + B image of FIG. 5A, to the saturation level of the A + B image. FIG. FIG. 5C is a diagram showing distortion when the BPF filter processing is performed on the A image and the B image in the saturation region of the pixel values of the A + B image in FIG. 5B.

  The distortion generation range is (n−1) pixels before and after the saturation region when the number of taps of the BPF is “n”. Therefore, if the focus detection calculation is performed in the distortion generation range, the expected defocus amount cannot be calculated. Based on such a problem, the detection signal generation unit 109 of the AF evaluation value calculation unit 300 generates a control signal for controlling the BPF processing of the filter unit 111.

  As illustrated in FIG. 3, the detection signal generation unit 109 includes a first saturation detection signal generation unit 301 and a second saturation detection signal generation unit 302. The first saturation detection signal generation unit 301 is an area (hereinafter referred to as “saturation”) in which the saturated pixel exists in the image sensor 102 when the input A + B signal 117 that is an imaging image signal reaches a preset saturation level. A first saturation detection signal 602 indicating a “pixel region” is generated. Then, the first saturation detection signal generation unit 301 outputs the generated first saturation detection signal 602 to the second saturation detection signal generation unit 302 and the NG determination unit 110.

  The second saturation detection signal generation unit 302 has a range in which distortion is generated in the phase difference image signal (delayed A signal 119 and B signal 120) from the input first saturation detection signal 602 due to the filter processing by the BPF. Determine. The second saturation detection signal generation unit 302 generates a second saturation detection signal 603 that suppresses the influence of distortion from the first saturation detection signal 602 and outputs the second saturation detection signal 603 to the selection selector unit 304 of the filter unit 111. .

  A method for generating the first saturation detection signal 602 and the second saturation detection signal 603 in the detection signal generation unit 109 will be described with reference to FIG. FIG. 6 is a timing chart of various signals when the BPF filter tap length is 5 taps. 6A shows the A + B signal 117 input to the detection signal generator 109, FIG. 6B shows the first saturation detection signal 602, and FIG. 6C shows the second saturation detection signal 603. Show. 6D shows the delayed A signal 119 input to the filter unit 111, FIG. 6E shows the BPF input signal 605 (see FIG. 3) input to the BPF 303A (see FIG. 3), and FIG. ) Shows the BPF output signal 606 (see FIG. 3) output from the BPF 303A. 6A, 6D, 6E, and 6F, the vertical axis indicates the pixel value, and the vertical axes in FIGS. 6B and 6C indicate "HIGH / LOW". .

  FIG. 6A shows that the A + B signal 117 that is an input signal to the detection signal generation unit 109 is saturated at the point T1, and the solid line between the points T1 and T2 indicates a saturated pixel region. The broken line indicates the pixel value in ideal A / D conversion. The first saturation detection signal 602 shown in FIG. 6B becomes HIGH at the point T1 when the A + B signal 117 reaches the saturation level, and becomes LOW at the point T2 where the saturation level changes from the saturation level to a level that maintains linearity. . That is, the first saturation detection signal 602 is HIGH in the saturated pixel region of the A + B signal 117, and is LOW in regions other than the saturated pixel region.

  The second saturation detection signal 603 shown in FIG. 6C becomes HIGH at the rising edge (point T1) of the first saturation detection signal 602, but the BPF with respect to the falling edge of the first saturation detection signal 602. It becomes LOW at the point T4 with a delay of 2 times in the range where the influence of the distortion appears. The range affected by this distortion is “(n−1) / 2” cycles when the number of BPF filter taps is “n”. In the example of the timing chart of this embodiment, since n = 5, the second saturation detection signal 603 changes to LOW after four cycles after the first saturation detection signal 602 falls. As described above, the second saturation detection signal 603 is obtained when the phase difference image signal is filtered by the BPF in a range in which the first saturation detection signal 602 becomes HIGH corresponding to the saturation pixel region of the imaging image signal. This signal is extended to a range where the effect of distortion is exerted.

  FIG. 6D shows that the delayed A signal 119 is saturated at the points T1 to T2. A BPF input signal 605 shown in FIG. 6 (e) is a signal obtained by processing the delay A signal 119 by the delay unit 305 (see FIG. 3), and is input to the BPF 303A with a delay of T1 to T3 that is affected by distortion. , Saturation occurs at the point T3. Due to this delay, the selection selector unit 304A (see FIG. 3) is applied based on the second saturation detection signal 603 to the distortion that occurs before and after the saturation start and end timing of the image signal for imaging (A + B signal 117). It becomes possible to control.

  FIG. 6 (f) shows that the BPF output signal 606 has a saturation region in the BPF input signal 605, so that it falls within the range from the rising edge to the falling edge of the second saturation detection signal 603 due to the influence of the filter processing by the BPF. , Indicates that it contains distortion. Thus, it can be seen that the second saturation detection signal 603 can avoid distortion.

  In FIG. 6, the delay A signal 119 input to the delay unit 305, the BPF input signal 605 generated by processing the delay A signal 119 by the delay unit 305, and the BPF input signal 605 by the BPF 303A are processed. The output BPF output signal 606 is shown. Similarly, based on the B signal 120 input to the delay unit 305, a BPF input signal to be input to the BPF 303B is generated in the delay unit 305, and the BPF 303B processes the BPF input signal to output the BPF output signal. Is output.

  The filter unit 111 that generates the BPF input signal and the BPF output signal includes BPFs 303A and 303B, selection selector units 304A and 304B, and a delay unit 305, as shown in FIG. The phase difference image signals (delayed A signal 119 and B signal 120) from the B image generation unit 113 are input to the delay unit 305. The delay unit 305 delays the input delay A signal 119 and B signal 120 according to the number of taps of the BPF 303A and 303B, respectively, and then outputs the delayed signal to the BPF 303A and 303B. As described above, the delay time is “(n−1) / 2” cycles when the number of taps is “n (natural number)”.

  FIG. 7 is a circuit diagram showing the configuration of the BPFs 303A and 303B. BPFs 303 </ b> A and 303 </ b> B are FIR filter circuits, and flip-flop (FF) circuits 701 to 704 having the same configuration, multipliers 705 to 709 for multiplying by filter coefficients, and addition for adding output signals from the FF circuits 701 to 704 And 710-713.

  The BPFs 303A and 303B are frequency removal means for removing a frequency unnecessary for focus detection (unnecessary in correlation calculation for detecting a phase difference) from the input phase difference image signal (BPF input signal). Is output to the selection selectors 304A and 304B. The outputs “y [n]” from the BPFs 303A and 303B are both “y [n] = k0 × x [n] + k1 × x [n−1] + k2 × x [n−2] + k3 × x [n−3]. ] + K4 × x [n−4] ″. Here, “x [n]” is input data, “k0 to k4” is a filter coefficient, “n” is a filter order (the number of multipliers, and in this embodiment, the same number as the number of taps of the BPFs 303A and 303B). ).

  FIG. 8 is a graph showing the filter characteristics of the BPFs 303A and 303B, with the horizontal axis representing the spatial frequency and the vertical axis representing the transmittance. As shown in FIG. 8, the BPFs 303A and 303B have characteristics of removing unnecessary frequencies in the correlation calculation (phase difference detection calculation) for detecting the phase difference outside the range of the specific cutoff frequencies f1 to f2. You can see that

The FF circuits 701 to 704 latch the phase difference image signal (BPF input signal) at the rising edge of the clock, and output it with a one cycle delay. In this embodiment, the delay amount when the respective inputs and outputs of the FF circuits 701 to 704 are connected in series is 5 cycles.

  Each of the multipliers 705 to 709 multiplies the input signal by the filter coefficients k0 to k4 and outputs the result. The filter coefficients k0 to k4 are set values that can extract an output signal corresponding to a frequency required for the focus detection calculation. For example, if the sampling frequency is “fs” and the cutoff frequencies are “f1” and “f2” (f1 <f2), the values “Ω1” and “Ω2” obtained by normalizing the cutoff frequency by the sampling frequency are “Ω1 = f1 / fs ”,“ Ω2 = f2 / fs ”. At this time, the filter coefficients of BPF 303A and 303B are “k0 = 2 (Ω2-Ω1)”, “kn = 2 (Ω2-Ω1) · cos (πn (Ω1 + Ω2)) · (sinπn (Ω2-Ω1)) / πn. (Ω2-Ω1) ”.

  The adders 710 to 713 process the output signals from the multipliers 705 to 709 as shown in FIG. 7 as follows. That is, the adder 710 adds the signal obtained by multiplying x [n] by k0 and the signal obtained by multiplying x [n] one cycle before by k1 and outputs the intermediate signal A1. The adder 711 adds the signal obtained by multiplying x [n] two cycles before by k2 and the intermediate signal A1, and outputs the intermediate signal A2. The adder 712 adds the signal obtained by multiplying x [n] three cycles before by k3 and the intermediate signal A2, and outputs the intermediate signal A3. The adder 713 adds the signal obtained by multiplying x [n] four cycles before by k4 and the intermediate signal A3, and outputs the output data y [n].

  The selection selector units 304A and 304B each output a phase difference image signal to the phase difference detection unit 107 based on the second saturation detection signal 603 generated by the detection signal generation unit 109. More specifically, the selection selectors 304A and 304B use, as input signals, phase difference image signals (BPF output signals) after processing by the BPFs 303A and 303B or signals having pixel values of zero (0) as mask signals. Based on the second saturation detection signal 603, one of these input signals is selected and output. At this time, the selection selectors 304A and 304B switch to the mask signal at the rising edge of the second saturation detection signal 603 and switch to the BPF output signal from the BPF 303A and 303B at the falling edge.

  The masking of the output signal by the input switching in the selection selector units 304A and 304B will be described with reference to FIG. In FIG. 9, the vertical axis represents the pixel value and the horizontal axis represents the horizontal scanning position, and the output waveforms (phase difference image signals) from the selection selector units 304A and 304B are indicated by solid lines. This output waveform (solid line) indicates that the input signal is switched to a mask signal having a pixel value of 0 instead of the BPF output signal in the selection selector units 304A and 304B at point G.

  Thus, in the output signal from the filter unit 111, the reference pixel value does not change before and after the input switching of the selection selector units 304A and 304B. Therefore, the occurrence of distortion of the phase difference image signals (delayed A signal 119 and B signal 120) due to the filter processing in the BPFs 303A and 303B is avoided with respect to the saturation of the imaging image signal (A + B signal 117). Can do.

  Based on the condition of the first saturation detection signal 602 input from the detection signal generation unit 109, the NG determination unit 110 sends a control signal indicating whether or not to perform a correlation calculation to the phase difference detection unit 107 and the AF control unit 106. Output. The condition of the first saturation detection signal 602 is, for example, the continuous period of the first saturation detection signal 602 or the cumulative number of saturated pixels per horizontal line. The NG determination unit 110 determines that the correlation calculation is not performed when a threshold set in advance in a ROM or the like (not shown) as a condition of the first saturation detection signal 602 is exceeded, and a determination indicating the determination result A signal (hereinafter referred to as “NG determination signal”) is generated.

  As a condition for generating an NG determination signal, a predetermined threshold “5” as a continuous period of a saturated pixel region is set as “100” as a predetermined threshold as a cumulative number of saturated pixels per horizontal line. The determination timing in the NG determination unit 110 will be described with reference to FIG. FIG. 11A shows the relationship between the first saturation detection signal 602 and the NG determination signal when five saturated pixels are consecutive. FIG. 11B shows the relationship between the first saturation detection signal 602 and the NG determination signal when there are 100 saturated pixels.

  In FIG. 11A, when the first saturation detection signal 602 is continuously input to the NG determination unit 110, the number of consecutive saturated pixels is counted by a counter (not shown), and five cycles are counted. Sometimes an NG determination signal is output. In FIG. 11B, when the first saturation detection signal is intermittently input to the NG determination unit 110, the cumulative number of saturated pixels is counted by a counter (not shown), and 100 pixels are counted. The NG determination signal is output. As described above, in each case of FIGS. 11A and 11B, when the count value exceeds the set number of pixels (threshold value), the NG determination unit 110 sends NG to the phase difference detection unit 107 and the AF control unit 106. Outputs a judgment signal.

  The phase difference detection unit 107 performs a correlation operation between the A signal and the B signal received from the filter unit 111 in accordance with the NG determination signal from the NG determination unit 110 and outputs the calculated defocus amount to the AF control unit 106. In addition, since the well-known technique according to the technique described in patent document 1 taken up as a prior art can be used for the method of calculating | requiring the defocus amount by a phase difference detection system, detailed description here is abbreviate | omitted. I decided to.

  As described above, according to the present embodiment, when the phase detection image signals (delayed A signal 119 and B signal 120) for focus detection are filtered by the BPFs 303A and 303B and the correlation calculation is performed, the divided PD 202 for the A image and the B image The influence of saturation in the divided PD 203 for use can be avoided. Thereby, accurate focus detection can be performed.

<Modification of AF Evaluation Value Calculation Unit>
A modification of the above-described AF evaluation value calculation unit 300 will be described with reference to the block diagram of FIG. The filter unit 111A included in the AF evaluation value calculation unit 300A according to the modification has a configuration in which gain units 1201A and 1201B are added as compared to the filter unit 111 of the AF evaluation value calculation unit 300.

  That is, the gain units 1201A and 1201B perform gain application processing for multiplying the phase difference image signals received from the BPFs 303A and 303B by a gain for reducing the pixel value level, and output the result to the selection selector units 304A and 304B. The selection selector units 304A and 304B select the phase difference image signals output from the gain units 1201A and 1201B and the phase difference image signals output from the BPFs 303A and 303B based on the second saturation detection signal 603, respectively. Switch to output. Here, the selection selectors 304A and 304B respectively output signals from the gain units 1201A and 1201B when the second saturation detection signal 603 rises, and output signals from the BPFs 303A and 303B when the second saturation detection signal 603 falls. Switch to the output signal.

  In the filter unit 111A, by switching the input in the selection selector units 304A and 304B, the amplitude around the saturation can be suppressed, and the degree of influence in the correlation calculation can be reduced. Thereby, the phase difference image signal with a small level difference. Can be output. Therefore, accurate focus detection can be performed.

<Other embodiments>
Although the present invention has been described in detail based on preferred embodiments thereof, the present invention is not limited to these specific embodiments, and various forms within the scope of the present invention are also included in the present invention. included. Furthermore, each embodiment mentioned above shows only one embodiment of this invention, and it is also possible to combine each embodiment suitably.

  For example, in the above-described embodiment, the PD 201 is divided into two parts, but may be a plurality of three or more divided parts. Further, the present invention is not limited to the band pass filter, but can be applied to filter processing using a high pass filter. Furthermore, the color filter 205 is normally configured by R, G, and B, but is not limited thereto, and may be configured using color filters such as cyan, magenta, and yellow as complementary color filters.

  In the above embodiment, the B image generation unit 113 generates the B signal. However, the present invention is not limited to this, and the B signal may be read to generate the A signal. In addition, the B image generation unit 113 generates the B signal not by the configuration having the delay line unit 114 but by the configuration having a buffer capable of storing one line or one frame, or the configuration connecting to a DRAM (not shown). It is good also as a structure.

  The NG determination unit 110 may be connected to the second saturation detection signal generation unit 302 so that the second saturation detection signal 603 is input. In that case, the influence range of BPF 303A and 303B is also set as a threshold value. That is, when the number of taps of the BPF 303A and 303B is “n” and the continuous number of saturated pixels is “5”, the threshold is “((n−1) / 2) +5” and the number of saturated pixels is “100”. The threshold value is “((n−1) / 2) +100”. With such a threshold setting, the NG determination unit 110 can determine whether or not to output an NG determination signal.

DESCRIPTION OF SYMBOLS 101 Lens group 102 Image pick-up element 106 AF control part 107 Phase difference detection part 109 Detection signal generation part 110 NG determination part 111 Filter part 202 Division | segmentation PD for A image
203 Divided PD for B image
300 AF evaluation value calculation unit 301 First saturation detection signal generation unit 302 Second saturation detection signal generation unit 303A, 303B BPF (band pass filter)
304A, 304B Selection selector section 1201A, 1201B Gain section

Claims (10)

An image sensor having a plurality of photoelectric conversion units sharing a microlens;
Generating means for generating an output signal from the plurality of photoelectric conversion units and an addition signal obtained by adding the output signals from the plurality of photoelectric conversion units;
Detecting means for detecting a saturated pixel region of the image sensor from the addition signal;
Frequency removing means for removing a predetermined frequency from output signals from the plurality of photoelectric conversion units;
Detection signal generation means for generating a saturation detection signal in accordance with the saturation pixel region detected by the detection means and the processing by the frequency removal means;
Selection means for outputting a signal processed by the frequency removing means or a mask signal as a phase difference image signal based on the saturation detection signal;
Defocus amount calculation means for calculating a defocus amount based on the phase difference image signal output from the selection means ,
The detection signal generation unit generates a first saturation detection signal indicating a saturated pixel region detected by the detection unit, and sets the range indicating the saturation pixel region in the first saturation detection signal as the plurality of photoelectric conversion units. Generating a second saturation detection signal expanded to a range in which the influence of the distortion is exerted when the output signal from is processed by the frequency removing means,
The selection means outputs the mask signal at the rising edge of the second saturation detection signal as the phase difference image signal, and the signal processed by the frequency removing means at the falling edge of the second saturation detection signal. A focus detection apparatus that performs switching so as to output .   The focus detection apparatus according to claim 1, wherein the frequency removing unit removes an unnecessary frequency in the correlation calculation for detecting a phase difference as the predetermined frequency. Based on the first saturation detection signal, comprising a determination means for determining whether to perform correlation calculation for focus detection,
The defocus amount calculation means includes a phase contrast image signal outputted from said selecting means, and Turkey to calculate the defocus amount of the optical system provided to the image pickup device based on the determination result by the determining means The focus detection apparatus according to claim 1 . The determination means determines that the correlation calculation for focus detection is not performed when a continuous period of the saturated pixel region indicated by the first saturation detection signal exceeds a predetermined threshold value. The focus detection apparatus according to claim 3 . The determination means determines that the correlation calculation for focus detection is not performed when the cumulative number of saturated pixels per horizontal line indicated by the first saturation detection signal exceeds a predetermined threshold. The focus detection apparatus according to claim 3 , wherein The frequency removing means is a bandpass filter;
The detection signal generation means generates the second saturation detection signal by expanding a range indicating the saturation pixel region in the first saturation detection signal in accordance with the number of taps of the bandpass filter. focus detecting apparatus according to any one of claims 1 to 5, characterized. The mask signal, a focus detection apparatus according to any one of claims 1 to 6, characterized in that a signal obtained by the pixel value to zero. As the mask signal, any of claims 1 to 6, further comprising a gain application processing means for generating a signal obtained by multiplying a gain to reduce the pixel value level for signals processed by the said frequency removing means The focus detection apparatus according to claim 1. The focus detection apparatus according to any one of claims 1 to 8 ,
A lens group that forms an optical image on an image sensor provided in the focus detection device;
Control means for performing drive control of the lens group based on a defocus amount output from a phase difference detection means provided in the focus detection device and a determination result by a determination signal provided in the focus detection device. An imaging device. A generation step of generating an output signal from the plurality of photoelectric conversion units of an imaging device having a plurality of photoelectric conversion units sharing a microlens and an addition signal obtained by adding the output signals from the plurality of photoelectric conversion units;
A detection step of detecting a saturated pixel region of the image sensor from the addition signal;
A frequency removing step of removing a predetermined frequency from the output signals from the plurality of photoelectric conversion units;
A detection signal generation step for generating a saturation detection signal in accordance with the saturation pixel region detected in the detection step and the processing in the frequency removal step;
A selection step of outputting the signal or mask signal processed in the frequency removal step based on the saturation detection signal as a phase difference image signal;
Have a, a defocus amount calculation step of calculating a defocus amount based on the phase contrast image signal outputted by the selection step,
The detection signal generation step generates a first saturation detection signal indicating the saturated pixel region detected in the detection step, and sets the range indicating the saturation pixel region in the first saturation detection signal to the plurality of photoelectric conversion units Generating a second saturation detection signal that is expanded to a range in which the influence of the distortion appears when the output signal from is processed in the frequency removal step,
The selection step outputs, as the phase difference image signal, the mask signal at the rising edge of the second saturation detection signal, and the signal processed in the frequency removal step at the falling edge of the second saturation detection signal. A focus detection method characterized by switching so as to output .

Images