JP3174124B2 - Focus detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a focus detection device for detecting a focus using a video signal, and more particularly to a focus detection device for detecting a focus by assigning frequency components.

[0002]

2. Description of the Related Art Japanese Patent Application Laid-Open No. 2-275916 discloses a method of detecting frequency components at two or more predetermined positions in an optical system and detecting a focal point by taking a ratio of the frequency components. ing. FIG. 13 shows an MD curve (frequency component: M
FIG. 14 is a diagram illustrating a relationship between a frequency component ratio and a defocus characteristic with respect to the MD curve in FIG. 13. From FIG. 14, it can be seen that the frequency component ratio and the defocus amount correspond to one to one. Hereinafter, the method of detecting the focal point will be briefly described.

The frequency signals at two locations having different optical path lengths are represented by M
S0a (ω), MS1a (ω), MTF
Assuming R, MS0a (ω) = O (ω) * L0 (ω) MS1a (ω) = O (ω) * L1 (ω) MTFR (ω) = MS1a (ω) / MS0a (ω) = L1 (ω) / L0 (ω) where O (ω); frequency component of subject L0 (ω), L1 (ω); frequency component of lens ω; frequency, *; multiplication.

As described above, the MTFR is a value that is not affected by the subject, and the MTFR has a one-to-one relationship with the defocus amount. Therefore, the defocus amount can be obtained without being affected by the subject.

In the discrete frequency detection (discrete Fourier transform), in order to prevent erroneous detection of a frequency due to a difference between signal values at both ends, a unique window function optimal for the system according to the system is used. Processing (with the center of the signal sequence as 1,
Weighting (weighting coefficient) is performed so that both ends are smoothly set to a value close to 0, and the detection frequency is optimized (theory of digital signal processing 1 Basic / System / Control Corona p77-) . Further, a method of setting a window function by focusing on signals at both ends of a signal sequence has been disclosed by the present applicant in Japanese Patent Application No. 4-10809.

[0006]

When the defocus amount is obtained from the above relational expression, a narrow-band frequency component is detected from the subject, and a false signal generated when the narrow-band is detected is surely canceled. There is a need.

As a method of detecting a narrow-band frequency component from an object image on a photoelectric conversion element, there is a method of detecting a frequency using an analog or digital filter or a method of detecting a frequency using a discrete Fourier transform. However, frequency detection from a limited area by a filter is very difficult to detect a narrow band signal. For example, in the case of an analog filter, it is difficult to set a constant number of each component, and in the case of a digital filter, the number of taps is very long, and it is difficult to extract a signal from a limited area.

On the other hand, the discrete Fourier transform cannot detect the original frequency component when there is a large difference between the signal values at both ends of the signal sequence in a limited area. Further, in order to remove the influence of the difference between the signal values at both ends, the signal values at both ends are set to 0.
Window processing is performed as follows. However, if window processing is performed uniformly, even if the signal difference between both ends is small, accurate frequency detection cannot be performed because the window processing is affected by adjacent frequency components on the frequency axis. p28-). Furthermore, when the window function is changed based on the signals at both ends of the signal sequence, the effect of convolution of the window function on the frequency axis cannot be completely removed.

The present invention has been made in view of the above-mentioned problems, and in order to detect an accurate frequency component, a high convolution coefficient for performing convolution processing on a signal sequence in accordance with signal information at both ends of the signal sequence is high. It is an object of the present invention to provide an in-focus detection device with high accuracy.

[0010]

That is, the present invention provides a photoelectric conversion element for converting a light distribution formed by the above-mentioned optical system into an electric signal at two places having different optical path lengths of the optical system, and a photoelectric conversion element comprising: Four to cut out a specific area
Cass area setting means and the focus area setting means
Set by a convolution coefficient setting means for setting a convolution coefficient based on the output from the photoelectric conversion element, set by the focus area setting means based on the convolution coefficient
Has been a convolution processing unit performs convolution processing on the output from the photoelectric conversion element, the convolution treated the follower
Set by Kasueria setting means, or photoelectric conversion element
Comprising a specific frequency detecting means for extracting a particular frequency from the output of al, a controller for performing focus adjustment of the optical system based on the characteristics of the specific frequency, and the focus area
Reduced output from photoelectric conversion element set by hand
Both on the basis of the difference value of the difference value and the primary differential value of the two ends signals, characterized in that so as to change the convolution coefficient.

[0011]

[Action] In the focus detection apparatus of the present invention, in the two positions having different optical path length of the optical system, the light distribution formed by the optical system is converted into an electric signal by a photoelectric conversion element, the photoelectric conversion Focus area setting means from element
Cuts out a specific area. Also, the above focus
The convolution coefficient is set by the convolution coefficient setting means based on the output from the photoelectric conversion element set by the swell setting means . The convolution processing means sets the focus area on the basis of the convolution coefficient.
A convolution process is performed on the output from the photoelectric conversion element, and the focus area setting after the convolution process is performed.
Set by means specific <br/> frequency withdrawn at a specific frequency detecting means from the output from the photoelectric conversion element. Focusing of the optical system is performed based on the characteristics of the particular frequency.
The convolution coefficient is determined by the focus area setting procedure.
At least both of the outputs from the photoelectric conversion elements
The change is made based on the difference value of the end signal and the difference value of the primary differential value.

[0012]

Embodiments of the present invention will be described below with reference to the drawings.

First, referring to FIG. 1, a first embodiment of the present invention will be described.
The concept as an embodiment of the present invention will be described. In FIG. 1, an in-focus point detecting device includes an optical system 11 and a photoelectric conversion element 1 for converting a light distribution of a subject passing through the optical system 11 into an electric distribution.
2, an A / D conversion circuit 13 for performing analog / digital conversion on the output of the photoelectric conversion element 12, a convolution coefficient setting circuit 14 for setting a convolution coefficient from the A / D converted signal, and a convolution coefficient setting circuit 14. Convolution (multiplication, addition and subtraction) of the coefficient set in the above with the A / D converted pixel signal
A convolution processing circuit 15 for processing; a frequency detection circuit 16 for detecting a plurality of frequency components from the convolution-processed pixel signal and storing the frequency components;
And a frequency component ratio detection circuit 17 for detecting a ratio of two frequency components by using the frequency components from No. 6.

Further, the in-focus point detecting apparatus is provided with a table 1
8, a configuration further including a defocus amount calculation circuit 19 and a drive circuit 20. The table 18 stores a relationship between each frequency component ratio obtained from a frequency component-defocus (MD) characteristic of the optical system 11 and a defocus amount. The defocus amount calculating circuit 19 calculates a defocus amount based on the frequency component ratio detected by the frequency component ratio detecting circuit 17 and a table value of the table 18 to change the state of the optical system 11. Then, the drive amount of the optical system 11 is calculated. Further, the driving circuit 2
0 is for driving the optical system 11 based on the defocus amount detected by the defocus amount calculation circuit 19.

In the in-focus point detecting device having such a configuration, when a light beam from a subject is guided by the optical system 11,
The light distribution of the subject is converted by the photoelectric conversion element 12 into an electric distribution. The pixel signal of the photoelectrically converted output is subjected to A / D conversion by an A / D conversion circuit, and then, by the convolution coefficient set by the convolution coefficient setting circuit 14,
The convolution processing is performed by the convolution processing circuit 15. A plurality of frequency components are detected and stored in the frequency detection circuit 16 of the convolution-processed pixel signal, and the two frequency component ratios are used to calculate the frequency component ratio using the frequency components.
7 is detected. Then, the defocus amount calculation circuit 19
In the above, the defocus amount is calculated based on the detected frequency component ratio and the table value of the table 18, and the optical system 11
Is detected, and the driving amount of the optical system 11 is calculated.
The drive circuit 20 drives the optical system 11 according to the drive amount.

That is, the focus detection apparatus focuses on the signals at both ends of the A / D converted focus area which cause the generation of an erroneous frequency component in the convolution coefficient setting circuit 14. The convolution coefficient is made variable in accordance with the difference between the signal value and the first-order differential value at both ends.

Here, a brief description of the method of focus detection will be given. Now, a signal sequence convolution process P (x) is performed,
The signal sequence is converted such that the first derivative and the absolute value are equal at both ends of the signal sequence. Since the convolution processing on the signal sequence is performed by multiplication on the frequency axis, if the frequency signals at two locations having different optical path lengths are MS0a (ω) and MS1a (ω), and the frequency component ratio is MTFR, MS0a (ω) = O (ω) * L0 (ω) * P (ω) MS1a (ω) = O (ω) * L1 (ω) * P (ω) MTFR (ω) = MS1a (ω) / MS0a (ω) = L1 ( ω) / L0 (ω) where O (ω); frequency component of subject P (ω); frequency component of convolution coefficient (convolution processing is set so that detection frequency ω does not become P (ω) 0) L0 ( ω), L1 (ω); frequency component of lens ω; frequency, *; multiplication. The effect of convolution is canceled for the detected MTFR (ω), and the defocus amount is obtained as described above. Next, a description will be given of a second embodiment in which the in-focus point detecting device according to the present invention is applied to a camera.

FIG. 2 is a diagram showing a schematic configuration of the camera. The light beam that has passed through the optical system 11 in the camera 21 is split into a finder system and an AF detection system by a main mirror (M mirror) 22 composed of a half mirror only in the central periphery. The light beam guided to the finder system is guided to the finder 23. The light beam guided to the AF detection system is guided to a beam splitter 25 by a sub-mirror (S mirror) 24, and is further converted to an electric signal by sensors 26a and 26b. In this sensor, the short side of the optical path length is 26a, and the long side of the optical path length is 26b. Further, a CPU 30 described later calculates a defocus amount of the optical system 11 based on the electric signal, and the optical system 11 is driven by the driving unit 27.

FIG. 3 is a block diagram of the camera shown in FIG. In FIG. 1, the camera includes a lens 28 and a sensor 29 for converting a light beam passing through the lens 28 into an electric signal.
A / D for converting a signal of the sensor 29 into a digital signal
The circuit 13, the CPU 30, and a detection frequency component D based on the pixel signal group and the detection frequency given by the CPU 30.
A DFT circuit 31 for detecting by FT processing (Discrete Fourier Transform) and a lens 2 for a plurality of frequency bands
A memory 32 for storing the relationship between the frequency component ratio of the lens 8 and the defocus amount of the lens 28;
And a lens drive circuit 20 for driving the lens 28 based on the amount of drive.

The CPU 30 includes the A / D conversion circuit 1
In 3, a convolution coefficient is determined based on the pixel information of the focus area that has been A / D converted, and convolution processing is performed using the convolution coefficient in the pixel signal. The calculated value and the detected frequency are sent to the DFT circuit 31 for detecting the frequency, and the necessary detected frequency is fetched from the DFT circuit 31 to calculate the frequency component ratio at two positions having different optical path differences. Then, in order to correct the magnification of the detected image and optimize the frequency, pupil information, frequency information, current focal position information, aperture value at the time of photographing, MTF information, focal length (zoom value) of the lens 28 are communicated, 29 is controlled. Further, the drive amount of the lens 28 is calculated by communication with the memory 32, and the drive amount is transmitted to the lens drive circuit 20.
Next, the operation of the camera thus configured will be described with reference to a flowchart as a main sequence in FIG.

When the operation is started, initialization is performed in step S1, and then it is determined in step S2 whether the first release is turned on. If the first release is off, this sequence is ended. If the first release is on, the process proceeds to step S3 to determine whether or not the second release is turned on. If the second release is ON, the process proceeds to step S4, and after the shutter and the winding sequence are performed,
It returns to step S2.

If it is determined in step S3 that the second release is off, the flow advances to step S5 to execute the focus detection subroutine program AF. Then, if the subroutine program AF has been executed, the focus flag GF is determined in step S6. Here, the flag GF indicates G when the lens 28 is at the in-focus position.
F = 1 and GF = 0 when the lens 28 needs to be driven out of focus. In step S6, if GF = 1, the process returns to step S2. On the other hand, when GF = 0,
Proceeding to step S7, the lens 28 is driven to the position designated by the subroutine program AF, and then returns to step S2. FIG. 5 shows a sequence of the subroutine program AF shown in FIG.

When the subroutine program AF is started, first, in step S11, information on the sensor corresponding to the focus area is initialized. Next, in step S12, integration of the sensor 29 is performed. After the integration is completed, the sensor signal is read in step S13, and the integrated signal is stored as digital data by A / D conversion. Thereafter, the process proceeds to step S14, the detection frequency is set to f _1. In addition, FIG.
As shown in, the detection frequency f _{1 is, f 10, f 11, f} 12 is selected according to the detected defocus range.

Next, in step S15, a subroutine program AFDC for pre-processing the digital signal of the pixel is performed. Then, in step S16, the detection frequency f ₁
, The detection frequency f ₁ is detected by digital processing. Further, in step S17, the ratio of the frequency components from the two sensors having a predetermined optical path difference (MTFR1) is detected, in the step S18, the MTFR1 the predetermined value epsilon ₁ are compared. Here, | MTFR1-1 | <
If not ε _1, the process proceeds to step S19, MTFR1
Detection Frequency defocus amount refers to the table table stored in f _1, the driving amount L1 interpolated by the lens 28 of the lens 28 is calculated as. Thereafter, in step S20, the drive amount L1 is stored in L0.

On the other hand, in step S18, | M
TFR1-1 | If <epsilon _1, the lens 28 is meant already be in the vicinity of the focal point. Therefore, the process proceeds to step S21, changes the detection frequency in order to increase the focus detection accuracy is performed, f ₂ (f ₂ of the frequency detection frequency is higher than f _1, the two sensors intervals shown in FIG. 6 (A frequency that is smaller than the lens frequency component and the defocus amount characteristic Z0).

Then, the detection frequency f _{2 is} detected in step S22 by the same digital processing (DFT) as in step S16. Then, step S2
3 The ratio of the frequency components from the two sensors having a predetermined optical path difference (MTRF2) is detected by, MTFR2 the predetermined value epsilon ₂ are compared in step S24.

In step S24, | MTFR2-
1 | If <not a ε _2, the process proceeds to step S25, M
Tables Table defocus amount is stored in the detection frequency f ₂ of the TFR2 and the lens 28 is referenced, it is interpolated, the driving amount L2 of the lens 28 is calculated. After the driving amount of the lens 28 is calculated in step S26 and step S20, the focus flag GF is determined in step S27.
Is set to GF = 0.

In step S24, | M
TFR2-1 | If <epsilon _2, focus flag GF is set to GF = 1 proceeds to step S28. Thus,
After the focusing flag GF is set, the subroutine program AF ends.

FIG. 6 shows frequency components at a specific frequency of the lens and defocus characteristics. In the figure, the vertical axis represents the frequency component of the lens (the focal point is normalized to 1 in an ideal subject), and the horizontal axis represents the defocus amount of the lens. The lens frequency component M0 is set in consideration of noise, and the defocus amount from the focal point at that time is set to Z0.

FIG. 7 shows a comparative value ε for MTFR.
₁ is a diagram for describing setting of epsilon _2. In the figure, the vertical axis represents the frequency component ratio, and the horizontal axis represents the defocus amount. The above ε ₁ becomes a switching point of the detection frequencies f ₁ and f ₂ and is determined in consideration of the difference between the two optical paths and the S / N of the signal. ε ₂ is determined in consideration of the focusing accuracy (determined by the F number of the lens) and the S / N of the signal. FIG. 8 is a subroutine showing a sequence of preprocessing of a digital signal of a pixel.

When the subroutine program AFDC is started, sensor data is read in step S31, and the front and rear sensor data are D (1, *), D
(2, *) (where * indicates 1 to 64 variables). Next, in step S32, the absolute values AS1 and AS2 of the difference values of the sensor data are obtained. And
In step S33, differential values BS11, BS12, BS21, and BS22 at both ends of each sensor of the sensor data are obtained.

Next, in step S34, the process of determining the differential value obtained in step S33 (BS11 = BS
12 and BS21 = BS22). Here, when the differential values match, it is determined that the sensor signal sequence that has been first-order differentiated at both ends of the sensor is continuous, and the process proceeds to step S35, where a difference value determination process (AS1 = 0 and AS2 = 0) is performed. If the difference values are equal in this step S35,
The process is continuous at both ends of the sensor, and the flow advances to step S36 to execute a data conversion subroutine program DCA.

On the other hand, if the difference values are not equal in step S35, the flow advances to step S37 to execute a subroutine program DCB for performing data conversion by convolution processing so that both ends of the sensor are continuous.

If the differential values at both ends of the sensor are not equal at step S34, the process proceeds to step S38, at which a subroutine program DCC for executing data conversion by convolution processing so that both ends of the sensor are continuous is executed. You. FIG. 9 shows a sequence of a subroutine program DCA for performing data conversion.

When the subroutine program DCA is started, first, in step S41, the average value AV of the signal train
1. AV2 is required. Next, in step S42,
The variable i is set to i = 1. In step S43, data conversion (subtraction of the average value from the original signal) of the pixel signal sequence is performed. Thereafter, in step S44, the determination of the variable i (i = 64) is performed.

Here, if i = 64 is not satisfied (all processing of the pixel signal sequence has not been completed yet), step S4
The process proceeds to 5, where i = i + 1 is performed, and the process returns to step S43. On the other hand, if i = 64 in step S44 (all the processes of the pixel signal sequence have been completed), the process proceeds to step S46, where MAX and MIN of all the pixel signal sequences are detected.

Next, in step S47, a normalization variable k (k = K / (MAX-MIN); where K is a predetermined standard value) is obtained. Then, in step S48, the variable i is set to i = 1. Thereafter, in step S49, the pixel signal sequence is normalized (the original signal is multiplied by k).
Is performed.

Next, in step S50, the variable i is determined (i = 64). Here, if i = 64 is not satisfied (all processing of the pixel signal sequence has not been completed yet), the process proceeds to step S51, where i = i + 1 is performed, and then returns to step S49. On the other hand, step S50
When i = 64 (all processing of the pixel signal sequence is completed)
Exits this sequence. FIG. 10 shows a sequence of a subroutine program DCB relating to the convolution processing when the changes at both ends of the pixel signal sequence are equal.

When the subroutine program DCB is started, in step S61, a new DB is set so that the convolution processing does not significantly affect the detection frequency.
Data of (1, 0) and D (2, 0) are temporarily created (the created data has continuity with D (1, 1) and D (2, 1), respectively). Next, in step S62, the variable i is set to i = 1. Further, step S6
In 3, convolution processing (convolution variables 1, -1)
Is performed on the pixel signal sequence.

Then, in step S64, the variable i
Is determined (i = 64). In step S64, if i = 64 is not satisfied (all processing of the pixel signal sequence has not been completed yet), the process proceeds to step S65, i = i + 1 is performed, and the process returns to step S63. On the other hand, if i = 64 in the above step S64 (all processing of the pixel signal sequence is completed), the process proceeds to step S66, where the subroutine program DCA is performed. Then, the process exits from this sequence. FIG. 11 shows a subroutine program DCC relating to convolution processing when the changes at both ends of the pixel signal sequence are not equal.

When the subroutine program DCC is started, first, in step S71, a new DDC is set so that the detection frequency is not significantly affected by the convolution processing.
(1,0) D (1,65), D (2,0), D (2,6)
5) Data is temporarily created (new data to be created are D (1,1), D (1,64), D (2,1), D
(2, 64). Next, in step S72, the variable i is set to i = 1. Then, in step S73, a convolution process (convolution variables 1, -2, 1) is performed on the pixel signal sequence. Then, in step S74, the determination of the variable i is performed (i = 64).

Here, if i = 64 is not satisfied (all processing of the pixel signal sequence has not been completed yet), step S7
5 and i = i + 1 is performed, and then step S7
Return to 3. On the other hand, if i = 64 in step S74 (all of the pixel signal sequence is completed), the process proceeds to the next step S76, where the subroutine program DCC is performed. afterwards,
Exit this sequence. Next, calculation of the drive amount of the lens 28 will be described with reference to the flowchart of FIG.

This drive amount is determined by the MT corresponding to the detection frequency.
Detection is performed using a table indicating the relationship between the FR and the defocus amount (the defocus points are discretely sampled). The detected MTFR is M ₁ , the defocus amount to be obtained is D _1, and the table MTF closest to M ₁ in the reference table.
The value of R is M ₂ , and the defocus amount corresponding to M ₂ is D
2. The value of the next table MTFR is M ₃ , and the defocus amount corresponding to M ₃ is D3.

When the driving amount calculation is started, step S8 is performed.
1 M ₂ can be obtained from the reference table at. Then, in step S82, M ₃ is obtained from the lookup table. Then, step S83 in _{D1 = D2 + | (M 2} -M 1) / (M 2 -M 3) | *
From (D3-D2), the defocus amount D1 of M ₁ is determined. After that, in step S84, the drive amount is calculated from the detected defocus amount and the current lens position, and the process exits this sequence.

With this configuration, it is possible to provide a focus detection device that can prevent the generation of spurious signals at both ends of the signal by the DFT processing with a high-speed and simple configuration and does not depend on the state of the subject. it can. In the embodiment described above,
Although DFT processing is used in frequency detection, any processing (digital filter, analog filter) that can detect a narrow band may be used. Although the detection frequency is switched twice this time, the detection of the frequency component may be performed in parallel, and the detection frequency may be switched a plurality of times. In addition, a plurality of detection frequencies may be used simultaneously within a range satisfying the condition of FIG. Although the defocus amount is obtained from the frequency component ratio using a table, it may be approximated by a simple function and stored in a counting memory.

Further, although the frequency is switched based on the frequency component ratio, the current position information of the lens may be used.
In addition, instead of the frequency component ratio, a value obtained by normalizing the difference between the two frequency components by the sum of the two frequency components may be used. Alternatively, the imaging optical system may be driven without using two sensors, and a signal may be captured in the next division. Note that preprocessing such as normalizing the light amount difference between the two sensor signals by the sum of the respective sensor signals may be performed. Further, the sensor signal is processed based on the average value, but may be based on the primary regression line.

Although the convolution coefficient has been processed up to the first-order differential coefficient, higher-order processing may be performed (used within a range where the value of the convolution processing does not become "0" at the detection frequency). .

Although the sensors are arranged at regular intervals with the film equivalent plane interposed therebetween, they need not always be arranged at regular intervals. In a system in which the front and rear defocus amounts are significantly different, it is better not to sandwich the film surface. Furthermore,
Any system that detects a relative frequency change in a system that detects a frequency by the DFT process can be used in other systems (such as a hill-climbing system).

[0049]

As described above, according to the present invention, the ratio of the change in the frequency of the subject due to the optical path difference is accurately detected in a narrow band from two video signals in different in-focus states. The frequency component of the subject can be canceled in a specific manner, and a highly accurate in-focus detection device can be provided with a simple configuration.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a concept as a first embodiment of a focus detection apparatus according to the present invention;

FIG. 2 is a diagram showing a schematic configuration of a camera in a second embodiment in which the in-focus point detection device of the present invention is applied to a camera.

FIG. 3 is a block diagram of the camera shown in FIG. 2;

FIG. 4 is a flowchart as a main sequence for explaining the operation of the camera in FIG. 2;

FIG. 5 is a sequence of a subroutine program AF shown in FIG. 4;

FIG. 6 is a diagram illustrating characteristics of a lens frequency component and a defocus amount.

FIG. 7 is a diagram for describing setting of comparative values ε ₁ and ε ₂ for MTFR.

FIG. 8 is a subroutine showing a sequence of preprocessing of a digital signal of a pixel in FIG. 5;

FIG. 9 shows a subroutine program DC for performing data conversion.
3 shows the sequence of A.

FIG. 10 shows a sequence of a subroutine program DCB relating to convolution processing when changes at both ends of a pixel signal sequence are equal.

FIG. 11 is a subroutine program DCC relating to convolution processing when changes at both ends of a pixel signal sequence are not equal;
It is shown.

FIG. 12 is a flowchart illustrating calculation of a drive amount of a lens.

FIG. 13 is a diagram showing an MD curve (a curve of a frequency component and a lens defocus characteristic) representing a general relationship between a frequency component and defocus.

FIG. 14 is a diagram showing a relationship between a frequency component ratio and a defocus characteristic with respect to the MD curve of FIG.

[Explanation of symbols]

11 optical system, 12 photoelectric conversion element, 13 A / D conversion circuit, 14 convolution coefficient setting circuit, 15 convolution processing circuit, 16 frequency detection circuit, 17 frequency component ratio detection circuit, 18 table 19: defocus amount calculation circuit, 20: drive circuit, 21: camera, 22: main mirror (M mirror), 23: finder, 24: sub mirror (S mirror), 25: beam splitter, 26a, 26
b: sensor, 27: drive unit.

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G02B 7/28 G02B 7/36 H04N 5/232

Claims (1)

(57) [Claims] 1. A photoelectric conversion element for converting a light distribution formed by the optical system into an electric signal at two places having different optical path lengths of the optical system, and a focus for cutting out a specific region from the photoelectric conversion element.
An area setting means, set by the focus area setting means, and convolution coefficient setting means sets a coefficient convolution on the basis of an output from the photoelectric conversion element, the focus area setting hand based on the convolution coefficient
Convolution processing means for performing convolution processing on the output from the photoelectric conversion element set in the stage, and the focus area setting means which has been subjected to the convolution processing
The specific frequency from the output from the photoelectric conversion element set in
A specific frequency detecting means for extracting the number, and a control means for adjusting the focus of the optical system based on the characteristic of the specific frequency , wherein the photoelectric element set by the focus area setting hand is provided.
Difference value between the at least two ends signals output from the child and on the basis of the difference value of the primary differential value, changing the convolution coefficients
An in- focus point detection device, characterized in that: