JP2014238452A - Automatic focus detection method and automatic focus detection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an automatic focus detection device and automatic focus detection method capable of reducing an effect of a noise component on focus detection accuracy.SOLUTION: An automatic focus detection method includes the steps of determining a difference signal value between one image signal value and the other corresponding to a pair of line sensors, performing square correlation operation determining the total sum of the square of the difference signal value to determine a square correlation function value of a pair of image signal strings, and determining a square correlation shift position having the highest similarity of the pair of image signal strings from the square correlation function value and determining a phase difference from the square correlation shift position.

Description

  The present invention relates to an automatic focus detection method for performing automatic focus adjustment control in a single-lens reflex camera or the like, and an automatic focus detection apparatus to which the automatic focus detection method is applied.

  A so-called phase difference type automatic focus detection apparatus projects a pair of pupil-divided subject images onto a line sensor and images a pair of subject images (a pair of image signal sequences output from the line sensor) on the line sensor. In this configuration, the defocus amount is obtained from the shift amount (the phase difference between the waveforms of the pair of image signal sequences). In the conventional phase difference detection method, a correlation calculation is performed on the pair of image signal sequences by a so-called image shift method, and a relative image shift between the pair of image signal sequences is performed according to the result of the correlation calculation. The amount is obtained, and the focus state (defocus amount) of the photographing lens is detected based on the image shift amount.

  Conventionally, this image shift amount is calculated by a linear three-point interpolation method (hereinafter referred to as “linear interpolation method”) as disclosed in, for example, paragraphs [0096] to [0100] of Patent Document 1. Alternatively, the amount of image shift at which the correlation amount is maximized is obtained with an accuracy equal to or less than the data sampling pitch (line sensor pixel pitch). As another method of three-point interpolation, a method of three-point interpolation (interpolation) with a quadratic function (hereinafter referred to as “second-order function interpolation method” or “second-order function interpolation operation”) is disclosed. (Paragraphs [0101] to [0104] of Patent Document 1). In the latter quadratic function interpolation operation, the case of performing the former linear interpolation operation (FIG. 5) becomes unnecessary.

JP 05-027162 A

  However, since the sum of absolute values of the difference signal values of the reference side output and the reference side output (linear correlation function value) is used as the correlation function, a noise component having a relatively small amplitude is also converted into a signal component having a relatively large amplitude. Similarly, since the correlation function is affected, the influence of the noise component on the image shift amount calculation variation (accuracy) is relatively large. Further, it has been found that the method of interpolating the linear correlation function value with a quadratic function at three points has poor adaptability (focus detection error and variation increase).

  In view of the above problems, an object of the present invention is to provide an automatic focus detection method and an automatic focus detection device that can reduce the influence of noise components on focus detection accuracy.

  The present invention has been completed by finding that the accuracy and accuracy of focus detection are improved by performing quadratic function interpolation when square correlation is calculated.

  That is, according to the present invention, a pair of subject images that have passed through different pupil positions of a photographing lens are projected onto a pair of line sensors having corresponding pixel columns, and pixels related to a pair of image signal sequences by the pair of line sensors. The correlation value is calculated by relatively shifting the signal value, the phase difference between the pair of image signal sequences is obtained based on the shift position where the correlation value is an extreme value, and the defocus amount of the photographing lens is calculated from the phase difference. In the phase difference type automatic focus detection method for obtaining the difference, a step of obtaining one and the other of the corresponding pixel signal values of the pair of line sensors, and a square correlation for obtaining a sum of squares of the difference signal values Performing a calculation to obtain a square correlation function value of the pair of image signal sequences; obtaining a square correlation shift position having the highest degree of matching between the pair of image signal sequences from the square correlation function value; Multiplicative correlation shift It is characterized by having a step of obtaining the phase difference from the position.

  The present invention relating to an automatic focus detection apparatus includes a focus detection optical system that projects a pair of subject images that have passed through different pupil positions of a photographing lens onto a pair of line sensors having corresponding pixel rows, and the pair of line sensors. To calculate a correlation amount by relatively shifting pixel signal values related to a pair of image signal sequences, and obtaining a phase difference between the pair of image signal sequences based on a shift position where the correlation amount is an extreme value. A focus detection calculation unit that obtains a defocus amount of the photographic lens from a phase difference, and a phase difference type automatic focus detection device, wherein the focus detection calculation unit includes one of the corresponding pixel signal values of the pair of line sensors. The other difference signal value is obtained, and a square correlation operation for obtaining the sum of the squares of the difference signal values is performed to obtain a square correlation function value of the pair of subject image signal sequences, and from the square correlation function value A pair of images No. obtains the highest square correlation shift position coincidence degree column is characterized by determining said phase difference from the square correlation shift position.

  In the automatic focus detection method of the present invention, the square correlation function may be a square correlation function value formed by a sum of squares of one of the pair of image signal sequences and the other difference signal value, and one of the pair of image signal sequences. It is practical to determine that the degree of coincidence between the pair of image signal sequences is higher as the square correlation function value is smaller. is there.

  The square correlation shift position is obtained by performing quadratic function interpolation using a plurality of square correlation function values including the minimum value of the square correlation function value, and the true minimum value of the square correlation function value is obtained. It is preferable to obtain the obtained shift position.

  In the automatic focus detection method of the present invention, before the step of obtaining the square correlation function value, a linear correlation function value consisting of a sum of difference signal values of one of the pair of image signal sequences and the other is obtained. A step of performing linear correlation calculation obtained at a plurality of shift positions by sequentially shifting the other of one pair of image signal sequences and detecting a minimum value from linear correlation function values obtained at a plurality of shift positions In the step of calculating the square correlation including the step, it is preferable that the square correlation function value is obtained by performing the square correlation calculation at a predetermined number of shift positions including the shift position where the minimum value is obtained. .

  In the step of obtaining the square correlation function value, a linear correlation function value composed of the sum of one of the pair of image signal sequences and the other difference signal value is sequentially shifted with respect to one of the pair of image signal sequences. A step of performing a linear correlation calculation obtained at a plurality of shift positions, and after this step, a step of detecting a minimum value from the linear correlation function values obtained at a plurality of shift positions, and a calculation by the linear correlation calculation. Determining whether or not the square correlation function value is a minimum value at the shift position where the minimum value of the linear correlation function value is obtained, and determining that the square correlation function value is a minimum value, A step of obtaining a square function shift position at which a minimum value is obtained, and if it is not determined that the value is a minimum value, a predetermined number of shift positions including the shift position at which the minimum value of the linear correlation function value is obtained are again determined. It is preferred to have the steps of obtaining and performing a multiply-correlation calculation, the square function shift position minimum value is obtained for the square correlation function value.

  The predetermined number when the square correlation function value is obtained at a predetermined number of shift positions including the shift position where the minimum value of the detected linear correlation function value is obtained is the quadratic function interpolation calculation or the linear correlation interpolation calculation. It is preferable that the number is larger than the number of the plurality of shift positions used in the above.

  The automatic focus detection method of the present invention further includes a step of detecting a minimum value of the square correlation function value after the step of obtaining the square correlation function value, and a minimum value of the detected square correlation function value. It is preferable to include a step of performing a quadratic function interpolation operation using a predetermined number of square correlation function values and obtaining a shift position at which a true minimum value of the square correlation function value is obtained.

  The automatic focus detection method of the present invention further includes a step of detecting a minimum value of the square correlation function value obtained by the square correlation calculation after the step of obtaining the square correlation function value, and whether the minimum value has been detected. When determining whether or not the minimum value has been detected, performing a quadratic function interpolation operation using a predetermined number of square correlation function values including the minimum value of the detected square correlation function value, A step of obtaining a shift position at which a true minimum value of the square correlation function value is obtained, and when it is determined that the minimum value of the square correlation function value cannot be detected, a predetermined number including the minimum value of the linear correlation function value A step of performing a linear interpolation operation using the linear correlation function value to obtain a shift position where a true minimum value of the linear correlation function value is obtained may be included.

  In the automatic focus detection apparatus of the present invention, the focus detection calculation means further includes a linear correlation function value consisting of a sum of difference signal values of one of the pair of image signal sequences and the other before obtaining the square correlation function value. The linear correlation calculation is performed at a plurality of shift positions by sequentially shifting the other of the pair of image signal sequences with respect to the other, and the minimum value is detected from the linear correlation function values obtained at the plurality of shift positions. In the square correlation calculation, it is preferable that the square correlation function value is obtained by performing the square correlation calculation at a predetermined number of shift positions including the shift position where the minimum value is obtained.

  In the automatic focus detection apparatus of the present invention, the focus detection calculation means further obtains the square correlation function value and calculates a linear correlation function value consisting of a sum of difference signal values of one and the other of the pair of image signal sequences. A linear correlation operation is performed at multiple shift positions by sequentially shifting the other of a pair of image signal sequences, and then a minimum value is detected from the linear correlation function values obtained at the multiple shift positions. When the square correlation function value is the minimum value at the shift position where the minimum value of the linear correlation function value obtained by the linear correlation calculation is obtained, the square that can obtain the minimum value of the square correlation function value When the function shift position is obtained and the minimum value is not the minimum value, the square correlation calculation is performed again for a predetermined number of shift positions including the shift position at which the minimum value of the linear correlation function value is obtained. The local minimum of It is preferable to obtain the square function shift position is.

  The predetermined number when the square correlation function value is obtained at a predetermined number of shift positions including the shift position where the minimum value of the detected linear correlation function value is obtained is the quadratic function interpolation calculation or the linear correlation interpolation calculation. It is preferable that the number is larger than the number of the plurality of shift positions used in the above.

  In the automatic focus detection apparatus of the present invention, the focus detection calculation means further obtains the square correlation function value, detects a minimum value of the square correlation function value, and calculates the detected square correlation function value. It is preferable to perform a quadratic function interpolation operation using a predetermined number of square correlation function values including the minimum value, and to obtain a shift position where a true minimum value of the square correlation function value is obtained.

  In the automatic focus detection apparatus of the present invention, when the focus detection calculation unit further detects the minimum value of the square correlation function value obtained by the square correlation calculation after obtaining the square correlation function value, A quadratic function interpolation is performed using a predetermined number of square correlation function values including the detected minimum value of the square correlation function value to obtain a shift position at which a true minimum value of the square correlation function value is obtained. When the minimum value of the square correlation function value cannot be detected, linear interpolation is performed using a predetermined number of linear correlation function values including the minimum value of the linear correlation function value, and the true value of the linear correlation function value is calculated. The shift position where the minimum value of can be obtained may be obtained.

According to the present invention, the noise component having a relatively small amplitude is suppressed by squaring the difference signal value between the pixel signal values of the pair of image signal sequences, and the signal component having a large amplitude is emphasized. Variations in the detection value can be reduced.
According to the fourth aspect of the present invention, the minimum value of the linear correlation function value is first detected using the linear correlation function, and the square correlation is obtained at a predetermined number of shift positions including the shift position where the minimum value is obtained. Since the calculation is performed, the number of times of the square correlation calculation is small and the calculation time can be shortened.

It is a block diagram which shows the main structures of the single-lens reflex camera carrying the automatic focus detection apparatus of this invention. It is a schematic plan view which shows the focus detection area of AF module of the automatic focus detection apparatus. It is a model top view which shows the structure of the line sensor of AF module of the automatic focus detection apparatus. It is a figure explaining the correlation calculation using the line sensor of the same AF module. (A), (B) is a graph explaining a linear correlation function and a linear interpolation method. It is a graph explaining a square correlation function and a quadratic function interpolation method. It is a basic flow chart of a 1st embodiment of automatic focus detection operation of the present invention. It is a flowchart of the 1st Example of the square system interpolation calculation by the quadratic function interpolation method using the square correlation function value of this invention. It is a basic flow chart of a 2nd embodiment of automatic focus detection operation of the present invention. It is a flowchart of the 2nd Example of the square system interpolation calculation by the quadratic function interpolation method using the square correlation function value of this invention. It is a flowchart of the 3rd Example of the square system interpolation calculation by the quadratic function interpolation method using the square correlation function value of this invention. It is a graph which compares the dispersion | variation in the defocus amount of this invention and a prior art. It is a graph which compares the suitability of a linear correlation function and an interpolation method. It is a graph which compares the suitability of a square correlation function and an interpolation method. It is a graph which shows an example in case the bottom position (shift position of a minimum value) of a linear correlation function and a square correlation function differs.

  FIG. 1 is an embodiment in which the present invention is applied to an automatic focus detection apparatus for an AF single-lens reflex camera, and its main configuration is a block diagram. This AF single-lens reflex camera includes a camera body 11 incorporating an AF module (automatic focus detection module) 60 as an automatic focus detection device, and an AF compatible photographing lens 51 that can be attached to and detached from the camera body 11.

  The camera body 11 includes a body CPU 31 that comprehensively controls the camera body 11 and the photographing lens 51 and also operates as a correlation calculation unit, a repeated pattern determination unit, a contrast detection unit, and a selection unit.

  On the other hand, the photographing lens 51 includes a lens CPU 57 that controls the lens function, a gear block 53 that drives the focus adjustment lens 52 in the optical axis direction, and a joint 35 of the camera body 11 that is provided in the mount portion of the photographing lens 51. And a joint 55 that is detachably connected. The lens CPU 57 is connected to the peripheral control circuit 21 of the camera body 11 via the connection of the electrical contact groups 56 and 36, and the open and maximum F value information is connected to the body CPU 31 via the peripheral control circuit 21. Predetermined data communication for communicating focal length information, lens position (distance) information, and the like is performed.

  Most of the subject luminous flux that has entered the camera body 11 from the photographic lens 51 is reflected by the main mirror 13 toward the pentaprism 17 constituting the finder optical system, is reflected by the pentaprism 17 and exits from the eyepiece. Part of the subject light beam emitted from the pentaprism 17 is incident on the light receiving element of the photometry IC 18. On the other hand, a part of the light beam incident on the half mirror portion 14 formed at the center portion of the main mirror 13 is transmitted through the half mirror portion 14 and reflected downward by the sub mirror 15 provided on the back surface of the main mirror 13, and AF The light enters the module 60.

  The photometry IC 18 outputs an electrical signal photoelectrically converted according to the amount of received light as a photometry signal to the body CPU 31 via the peripheral control circuit 21. The body CPU 31 performs a predetermined exposure calculation based on a photometric signal or the like, and calculates an appropriate shutter speed and aperture value for exposure. During the photographing process, the peripheral control circuit 21 drives the mirror motor 25 via the mirror motor driver 24 to raise the main mirror 13 and also drives the diaphragm mechanism 22 to stop the diaphragm (not shown). Z) is set to the calculated aperture value, and exposure is performed by driving the exposure mechanism 23 based on the calculated shutter speed. Further, after the exposure is completed, the peripheral control circuit 21 drives the mirror motor 25 via the mirror motor driver 24 to lower the main mirror 13.

  The body CPU 31 includes a ROM 31a that stores a control program, a predetermined data for calculation and control, a RAM 31b that temporarily stores variables, a timer 31c for timing, a counter 31d, and an output VOUT signal ( A / D converter 31e for A / D conversion of the video signal) and a D / A converter 31f for D / A converting the monitor reference VMS signal for setting the end of integration of the AF module 60 and outputting it to the AF module 60 doing.

  The AF module 60 is a so-called pupil division phase difference method, and includes a focus detection element 61 having a CCD line sensor and a monitor sensor for monitoring the integration end timing, and a focus detection surface (not shown) equivalent to an imaging surface. And an AF optical system that divides the subject luminous flux forming the subject image into two and divides it into a corresponding CCD line sensor.

  The AF module 60 detects the focus state of the subject and outputs a pixel-unit video signal (a pair of image signal sequences) to the body CPU 31. The body CPU 31 converts the video signal input from the AF module 60 into a digital signal by the built-in A / D converter 31e, performs a predetermined calculation based on the digital signal corresponding to the focus detection area, and calculates the defocus amount. To do. Further, the body CPU 35 determines the rotation direction of the AF motor 33 based on the calculated defocus amount, calculates the rotation number as the number of AF pulses output from the encoder 37 that detects the rotation number of the AF motor 33, and is built-in. Set to the counter 31d. The body CPU 31 drives the AF motor 33 via the AF motor driver 32 based on the rotation direction and the number of pulses. During this driving, the body CPU 31 counts down the pulses output from the encoder 37 in conjunction with the rotation of the AF motor 33 by the built-in counter 31d, and stops the AF motor 33 when the count value becomes zero. The rotation of the AF motor 33 is decelerated by the gear block 34, and the gear of the photographic lens 51 is connected via a connection between a joint 35 provided on the mount portion of the camera body 11 and a joint 55 provided on the mount portion of the photographic lens 51. The focus adjustment lens 52 is transmitted to the block 54 and moved forward and backward in the optical axis direction via the gear block 54.

  The body CPU 31 has a focus switch SWAF for switching the focus mode between a manual mode and an AF (one-shot / continuous AF) mode, a photometry switch SWS that is turned on by half-pressing a manual release button (not shown), and a full-press. A main switch SWM for turning on / off the power to the release switch SWR and the peripheral control circuit 21 is connected.

  The body CPU 31 has a display panel 39 for displaying various shooting information such as a set AF, exposure, shooting mode, shutter speed, aperture value, and various constants specific to the camera body 11 as external nonvolatile memory means. A stored flash memory 38 is connected. The display panel 39 includes displays provided at two locations within the outer surface of the camera body 11 and the viewfinder field.

  The camera body 11 is provided with an image sensor 45 as imaging means. The imaging surface by the image sensor 45 and the focus detection surface by the AF module 60 are set to be equivalent. The output signal of the image sensor 45 is digitized by an AFE (analog front end) 46 and processed into a video signal that can be displayed on a display (LCD) 42 by a DSP 41. The DSP 41 exchanges information regarding photographing with the body CPU 31.

  FIG. 2 is a schematic plan view for explaining a focus detection area of the focus detection element 61. The focus detection element 61 is a CCD focus detection element having a line sensor IH onto which a pair of subject image light divided into pupils is projected. In the present embodiment, a focus detection area CC is set in the center of the shooting screen 70, and the subject image in the cross focus detection area CC is divided into a pair of subject images that pass through different pupil positions of the shooting lens. Thus, the image is projected onto different one and other regions of the corresponding line sensor IH.

  The line sensor IH is divided into two areas of a reference line sensor A and a reference line sensor B (see FIG. 4). FIG. 3 shows a reference line sensor A corresponding to the reference line sensor area of the line sensor IH. An example is shown. The line sensor IH is a CCD line sensor in which a total number N (N is a positive integer) photoelectric conversion elements are arranged in a horizontal direction (left and right in the figure, horizontal direction), as shown in FIG. Pixel numbers n = 1 to N are assigned to the pixels. The reference line sensor B of the line sensor IH has the same configuration as the reference line sensor A of FIG.

  One and the other of the pair of subject images divided into pupils are projected onto the standard line sensor A and the reference line sensor B on the line sensor IH, and the pixels (photoelectric conversion elements) in the respective line sensor regions are integrated. That is, it is photoelectrically converted into an electric charge corresponding to the illuminance of the subject image and accumulated as an electrical signal. When the integration is completed, the line sensor IH sequentially converts the integrated charges into a predetermined pixel signal, and outputs each as a pair of image signal sequences.

  When the body CPU 31 receives an image signal sequence (video signal) from the AF module 60, the body CPU 31 obtains a linear correlation function SCO (i) for the pair of image signal sequences from the line sensor IH by an image shift method. The correlation function SCO (i) indicates the degree of coincidence between the reference line sensor A and the reference line sensor B of the line sensor IH, and the difference (difference) between the pixel signal values of the image signal sequences of the reference line sensor A and the reference line sensor B. The correlation function SCO (i), which is the sum of the absolute values of the signal values), is obtained by shifting the reference line sensor B by one pixel from the reference line sensor A as a reference. The correlation function SCO (i) is a function that becomes smaller as the coincidence degree (waveform coincidence degree) between a pair of image signal sequences is higher. FIG. 4A shows a state in which the reference line sensor A of the line sensor IH and the reference line sensor B (the phase of the pixel signal sequence thereof) match with the shift number I = 0 in the initial state. If the maximum shift number is ± imax, the reference line sensor B is shifted by ± 1 pixel with respect to the reference line sensor A from the initial state of FIG. 4A as shown in FIGS. 4B and 4C. However, the calculation is repeated until the shift number I = ± imax from the specific shift position. 4D and 4E show the state where the shift number I = ± i. A state where the reference line sensor A of the line sensor IH and the reference line sensor B (the phase of the pixel signal sequence thereof) coincide with each other at the specific shift position is a focused state.

Ａ_ｎは基準ラインセンサＡの画素番号に対応する画像信号、Ｂ_ｎ＋ｉは参照ラインセンサＢの画素番号に対応する画像信号である。 The linear correlation function SCO (i) can be expressed by the following equation (1).

_An is an image signal corresponding to the pixel number of the reference line sensor A, and B _{n + i} is an image signal corresponding to the pixel number of the reference line sensor B.

  The phase difference (deviation amount) between the pair of subject images is a specific shift position (i = when the images of the reference line sensor A and the reference line sensor B (the waveforms of the image signal sequences IH-A and IH-B) match). The number of shifts (shift amount) from 0). Therefore, the body CPU 31 detects the smallest minimum value of the linear correlation function SCO (i), and determines the shift number of the image of the reference line sensor B with respect to the reference line sensor A when the minimum value is taken as the image shift amount (shifted). Detect as quantity). Note that the image correlation amount (shift amount) can be obtained only from the linear correlation function SCO (i) in units of the pixel (photoelectric conversion element) pitch of the line sensor IH. An interpolation operation is performed on (i) to obtain an accurate minimum value with small variations (high accuracy). The defocus amount of the photographing lens 51 is obtained from the number of shifts (difference between the shift position and the specific shift position) when the minimum value is obtained.

FIG. 5 shows an embodiment of the linear interpolation method (linear interpolation method). The linear interpolation method is a method for obtaining a true minimum value of a correlation function that may exist between shift positions, and a straight line passing through two correlation function values including a correlation function value that is a minimum value; In this method, the true minimum value of the correlation function is obtained from the intersection of the straight lines passing through the other correlation function values and having the opposite direction to the straight line. FIG. 5 shows the correlation function values H ₋₁ and H ₊₁ before and after sandwiching the correlation function value H ₀ and the correlation function value H ₀ as the minimum values from the correlation function values calculated by the linear correlation function SCO (i). these correlation function value H _0, H _-1, by applying a linear interpolation (linear interpolation) calculation at H ₊₁ illustrates graphically the manner of obtaining the true image shift amount. FIG. 5 (A), the case where the correlation function value H ₊₁ of the right is not less than the correlation function value H _-1 the left in the drawing, FIG. 5 (B), the correlation function value H ₊₁ right in figure left a graph showing a case of less than the correlation function value H _-1 of the horizontal axis is the number of shifts, and the vertical axis represents the correlation function value.

In the case of H _{+ 1} ≧ H− ₁ , the shift position of the straight line passing through the correlation function values H ₀ and H ₊₁ and the intersection of the straight line passing through the correlation function value H ₋₁ and the straight line passing through the correlation function value H ₋₁ are reversed. Ask. The shift position of this intersection is a true shift position or a shift position closer to the true shift position. Assuming that the difference between the shift position of the correlation function value H ₀ , which is the minimum value, and the shift position of the intersection point is the shift error β, the shift error β can be obtained by the following equation (2) (FIG. 5A).

In the case of H ₋₁ > H ₊₁ , the shift position of the intersection of the straight line passing through the correlation function values H ₋₁ and H ₀ and the straight line passing through the correlation function value H _{+1 in} which the straight line and the inclination direction are opposite is obtained. . The shift position of this intersection is a true shift position or a shift position closer to the true shift position. Assuming that the difference between the shift position of the correlation function value H ₀ that is the minimum value and the shift position of the intersection is the shift error β, the shift error β can be obtained by the following equation (3) (FIG. 5B).

By the above linear interpolation calculation, the image shift amount i ₀ + β that is a true shift position or a shift position closer to the true shift position can be obtained. The image shift amount i ₀ + β is more accurate and accurate than the image shift amount i ₀ obtained only by the correlation function.

FIG. 6 shows an embodiment of a quadratic function interpolation method (parabolic interpolation method). The quadratic function interpolation method of this embodiment exists between square correlation shift positions (bits) by a quadratic function (parabola) that can obtain three or more correlation function values including a correlation function value that is a minimum value. This is a technique for obtaining a true minimum value (extreme value) of a correlation function that may be performed. FIG. 6 shows the square correlation before and after sandwiching the square correlation function value H ₀ ^S and the square correlation function value H ₀ ^S , which are minimum values, from the square correlation function value calculated by the square correlation function SCO2 (i). The function values H ₋₁ ^S and H ₊₁ ^S are obtained, and a quadratic function interpolation method (quadratic function interpolation operation) is applied to these square correlation function values H ₀ ^S , H ₋₁ ^S and H ₊₁ ^S. FIG. 6 is a graph showing how to obtain a true minimum value, that is, an accurate phase difference (image shift amount).

In this embodiment, in order to obtain the degree of coincidence between the image signal sequences IH-A and IH-B of the standard line sensor A and the reference line sensor B, the square correlation function SCO2 (i) is obtained by the following equation (4).

In the above-described square correlation function SCO2 (i), the square correlation function value H ₀ ^S that is the minimum value and the square correlation function values H ₋₁ ^S and H ₊₁ before and after the square correlation function value H ₀ ^S are sandwiched. ^A square that obtains ^S and further applies a quadratic function interpolation operation to the three-point square correlation function values H ₀ ^S , H ₋₁ ^S , and H ₊₁ ^S to obtain a more accurate and accurate minimum value The correlation shift position is obtained. If the difference between the minimum square correlation function value H ₀ ^S and the shift position at which the true minimum square correlation function value is obtained is β ^S , the shift error β ^S can be obtained by the following equation (5). Yes (Fig. 6).

The square correlation function SCO2 (i) is the sum of the squares of the differences (difference signal values) between the pixel signal values of the image signal sequences of the standard line sensor A and the reference line sensor B, as is apparent from Equation (4). . The square correlation function SCO2 (i) suppresses a noise component having a relatively small amplitude and emphasizes a signal component having a relatively large amplitude, thereby reducing variation in image shift amount due to noise (inhibiting a decrease in accuracy). It becomes possible to do.
In the embodiment of the present invention, a quadratic function interpolation operation is performed on the calculated square correlation function SCO2 (i), so that an accurate image shift amount can be detected with high accuracy.

  FIG. 12 shows a phase difference method for a plurality of general subjects with a combination of a linear correlation function and a linear interpolation operation, and a combination of a square correlation function and a quadratic function interpolation. It is a graph which shows the result of having compared the dispersion | variation (accuracy) of the defocus amount when performing focus detection. The horizontal axis represents the defocus amount variation (a.u.) obtained by the linear interpolation operation, and the vertical axis represents the defocus amount variation (a.u.) obtained by the quadratic function interpolation operation. The horizontal axis and the vertical axis are shown in arbitrary units (a.u.) of the same scale. Also from this graph, it can be seen that it is preferable to perform focus detection by a combination of a square correlation function and a quadratic function interpolation because the variation in defocus amount is relatively small.

  FIG. 13 shows the shift position of the minimum value when linear interpolation (linear interpolation) calculation is performed on the linear correlation function SCO (i) and when quadratic function interpolation (secondary function interpolation) calculation is performed. It is a graph which compares (bottom position in a bit) and a focus detection error (defocus amount error). In the graph, the rhombus indicates the linear interpolation calculation value, the square indicates the quadratic function interpolation calculation value, and the horizontal axis indicates the shift position of the minimum value within the pixel 1 pitch (1 bit of the image signal sequence) (bottom in bit). Position (bit)), and the vertical axis indicates the focus detection error (au). From FIG. 13, it can be seen that for the linear correlation function SCO (i), the focus detection error is smaller when the linear interpolation is calculated than when the quadratic function is calculated.

  FIG. 14 shows the shift of the minimum value when linear interpolation (linear interpolation) calculation is performed on the square correlation function SCO2 (i) and when quadratic function interpolation (secondary function interpolation) calculation is performed. It is a graph which compares a position (bottom position in a bit) and a focus detection error. In the graph, the rhombus indicates the linear interpolation calculation value, the square indicates the quadratic function interpolation calculation value, the broken line indicates the linear correlation function value, the solid line indicates the square correlation function value, and the horizontal axis indicates the pixel 1 pitch ( The shift position (bottom position (bit) in the bit) of the minimum value in one bit of the image signal sequence, and the vertical axis indicates the focus detection error (au). As can be seen from FIG. 14, for the square correlation function SCO2 (i), the focus detection error is smaller when the quadratic function interpolation is performed than when the linear interpolation is performed.

  From the above, it is more accurate and accurate for linear correlation function SCO (i), and more accurate and accurate for quadratic function SCO2 (i). It turns out that it is. Therefore, the automatic focus detection apparatus of the present invention performs linear interpolation on the linear correlation function SCO (i) and performs quadratic function interpolation on the square correlation function SCO2 (i) under predetermined conditions. To do.

  The automatic focus detection operation of the present invention will be described in more detail with reference to the flowcharts shown in FIGS. FIG. 7 is a main flowchart showing an embodiment of the focus detection operation of the automatic focus detection apparatus of the present invention. This focus detection main flowchart is entered when the main switch SWM is turned on and the photometric switch SWS is turned on.

  When the automatic focus detection operation is started and automatic focus detection 1 is entered, a pair of image signal sequences acquired from the line sensor IH is passed through a low-pass filter to remove high-frequency noise components (S101), and the image passed through the low-pass filter. The contrast is calculated for the signal sequence (at least the image signal sequence of the reference line sensor A) (S103). It is checked whether or not the contrast is OK (S105). If it is determined that the contrast is not OK (S105: NO), the focus detection operation (automatic focus detection 1) is terminated (END).

  Next, when it is determined that the contrast is OK (S105: YES), the pair of image signal sequences are passed through a high-pass filter to remove DC noise components (S107).

For the pair of image signal sequences that have passed through the high-pass filter, linear correlation calculation is performed by an image shift method to obtain a linear correlation function value (S109). Next, in the obtained linear correlation function value, a minimal correlation function value H ₀ and a bottom position (shift amount) i _{0 from} which the correlation function value H ₀ is obtained are detected (searched), and the detected bottom position i is detected. ₀ and three correlation function values H ₀ , H ₋₁ , H ₊₁ (see FIGS. 5A and 5B) sandwiching the bottom position i ₀ are stored, and the process proceeds to step S113 (S111, S113). In step S113, it calculates the reliability of the bottom position _{i 0.} The reliability of the bottom position i ₀ is the magnitude of the inclination angle of the straight line passing through the correlation function values H ₋₁ and H ₀ , the inclination angle of the straight line passing through the correlation function values H ₊₁ and H ₀ , and the correlation function value H _0. It is obtained by a known method such as comparison with a threshold value.

  When it is determined that the reliability is OK (Yes) (S115: YES), the square method bit interpolation operation 1 is performed to finish the automatic focus detection 1 (S117, END), and it is determined that the reliability is OK (Yes). If not (S115: NO), step S117 is skipped and the automatic focus detection 1 is terminated (END).

  When the automatic focus detection 1 is completed, the defocus amount is calculated based on the image shift amount by another AF lens driving thread (not shown), and the focus adjustment lens group 52 is moved to the in-focus position based on the defocus amount. The lens drive amount (rotation direction and rotation amount of the AF motor 33) necessary for the operation is calculated, and the AF motor 33 is driven via the motor drive circuit 32 based on the calculated lens drive amount, and the gear block 34 and the joint 35 are driven. , 55 and the gear block 54, the focus adjustment lens group 52 is moved to the in-focus position.

The square method bit interpolation calculation 1 performed in step S117 will be described in more detail with reference to the flowchart shown in FIG. In the square method bit interpolation operation 1, first, a shift position of ± 2 bits sandwiching the bottom position i ₀ and the bottom position i ₀ of the linear correlation function value detected in step S111, that is, the shift position i ₀ -2, i _{0 -1, i 0, i 0} + 1, i 0 for 5 points +2 conducted square correlation operation finds square correlation function SCO2 (i) (S201). Subsequently, the square correlation bottom position i ₀ ^S (see FIG. 6) of the square correlation function SCO2 (i) obtained by the square correlation calculation is detected (S203), and the process proceeds to step S205. Since the shift range in which the square correlation calculation is performed is narrowed down to five points, the calculation time is shortened compared with the case where the minimum value is obtained by calculating the square correlation for the entire shift range from the beginning.

Here, the reason why the square correlation function value is obtained at the shift position of ± 2 bits across the bottom position i ₀ will be described. FIG. 15 is a graph showing an example in which the bottom position (minimum value bottom position) differs between the linear correlation function SCO (i) and the square correlation function SCO2 (i) for the same subject. In the graph, the rhombus is the linear correlation function SCO (i), the square is the square correlation function SCO2 (i), the horizontal axis is the shift amount (bit), and the vertical axis is the linear correlation function value and the square correlation function value. It is. As shown in the example of this figure, in a situation where the bottom position is likely to fluctuate, such as when the true bottom position is near the center of the bit shift, the bottom position of the linear correlation function is This is because the bottom position of the square correlation function may be found by finding the outermost square phase correlation function value, because the bottom position of the square correlation function may not always match.

In step S205, it is checked whether or not the bottom of the square correlation function (square correlation bottom position i ₀ ^S ) is present (detected). When it is determined that the square correlation bottom position i ₀ ^{S is} present (S205: YES), quadratic function interpolation is performed using the three-point square correlation function values H ₋₁ ^S , H ₀ ^S, and H ₊₁ ^S. Then, a deviation amount β ^S corresponding to the error between the bottom position where the true minimum value of the secondary correlation function is obtained and the square correlation bottom position i ₀ ^S is calculated (S207), and further corresponds to the true bottom position. The image shift amount i ₀ ^S + β ^S is calculated and the process returns (S209, RET) (see FIG. 6). When it is not possible to determine that the square correlation bottom position i ₀ ^{S is} present (S205: NO), the linear correlation function values H ₋₁ , H ₀ and H ₊₁ of the linear correlation function values detected in step S111 are linear. An interpolation operation is performed to calculate a shift amount β between the bottom position where the true minimum value of the linear correlation function value is obtained and the bottom position i ₀ (S211), and further an image shift amount i corresponding to the true bottom position. ₀ + β is calculated and the process returns (S213, RET).

According to the automatic focus detection 1 of the first embodiment of the present invention, the square correlation calculation is performed in a range of ± 2 bits with respect to the bottom position obtained in advance by the linear correlation calculation. The calculation time is shortened compared with the case where the power correlation calculation is performed. When the square correlation bottom position i ₀ ^S of the square correlation function can be detected, the image shift amount i ₀ ^S + β ^S is obtained by quadratic function interpolation, so that the more accurate and accurate image shift amount i ₀ ^{S can be obtained.} the + beta ^S, can be obtained in a short time. When the square correlation bottom position i ₀ ^S of the square correlation function cannot be detected, linear interpolation is performed using the correlation function values H ₋₁ , H ₀ and H ₊₁ detected from the linear correlation values. Thus, it is possible to obtain an accurate image shift amount i ₀ + β with high accuracy.

  When the automatic focus detection 1 is completed, the defocus amount is calculated based on the image shift amount by another AF lens driving thread (not shown), and the focus adjustment lens group 52 is moved to the in-focus position based on the defocus amount. The lens driving amount (rotation direction and amount of rotation of the AF motor 33) necessary for the calculation is calculated, and the AF motor 33 is driven via the motor drive circuit 32 based on the calculated lens driving amount, and the gear block 34, The focus adjustment lens group 52 is moved to the in-focus position via the joints 35 and 55 and the gear block 54.

  FIG. 9 is a flowchart showing a second embodiment (automatic focus detection 2) regarding the automatic focus detection operation of the present invention. The automatic focus detection 2 is different from the automatic focus detection 1 of the first embodiment in that the linear correlation calculation and the square correlation calculation are simultaneously performed and then the bottom position of the linear correlation function value is detected. When automatic focus detection 2 is entered, a pair of image signal sequences acquired from the line sensor IH is passed through a low-pass filter (S301), and then contrast is calculated for at least the image signal sequence of the reference line sensor A (S303). Subsequently, it is checked whether or not the contrast is OK (S305). If it is determined that the contrast is not OK (S305: NO), the automatic focus detection operation (auto focus detection 2) is ended (END).

  When it is determined that the contrast is OK (S305: YES), the pair of image signal sequences are passed through the high-pass filter and the process proceeds to step S309 (S307, S309).

In step S309, a linear correlation function value and a square correlation function are simultaneously performed on the pair of image signal sequences by an image shift method to obtain a linear correlation function value and a square correlation function (S309). For the obtained linear correlation function value, a bottom position i ₀ where a minimum value is obtained is detected (searched), and a total of three correlation function values H ₀ , H ₋₁ , H across the bottom position i ₀ and the bottom position i ₀ are detected. ₊₁ (see FIGS. 5A and 5B) is stored, and the process proceeds to step S313 (S311 and S313). In step S313, it calculates the reliability of the bottom position _{i 0.} The reliability of the bottom position i ₀ depends on the inclination angle of the straight line passing through the correlation function values H ₋₁ and H ₀ and the inclination angle of the straight line passing through the correlation function values H ₊₁ and H ₀ or the magnitude of the correlation function value H _0. It is obtained by a known method such as comparison with a threshold value.

  When it is determined that the reliability is OK (Yes) (S315: YES), the square bit interpolation operation 2 is executed and the process returns (S317, RET), and when the reliability is not OK (Yes) (S315). : NO), skipping step S317 and returning (RET).

FIG. 10 is a flowchart showing details of the square method bit interpolation operation 2 performed in step S317. When the square method bit interpolation operation 2 is entered, first, whether or not the square correlation function value calculated in step S309 is the bottom at the bottom position i ₀ of the linear correlation function value (the square at the bottom position of the linear correlation function value). Whether or not the correlation function value is a minimum value is checked (S401, FIG. 15). When it is determined that it is not the bottom (S401: NO), in each of ± 2 shift positions i ₀ -2, i ₀ -1, i ₀ , i ₀ +1, i ₀ +2 including the bottom position i ₀ of the linear correlation function value A square correlation calculation is performed and the process proceeds to step S405 (S403, S405). When it is determined that the square correlation function value is the bottom at the bottom position i ₀ of the linear correlation function value (S401: YES), step S403 is skipped and the process proceeds to step S405.

In step S405, a square correlation bottom position i ₀ ^S of the square correlation function value is detected, and a three-point square correlation function value H including the square correlation function value H ₀ ^S of the square correlation bottom position i ₀ ^S is detected. ₀ ^S , H ₋₁ ^S , H ₊₁ ^S (see FIG. 6) are obtained and stored, and the process proceeds to step S407.

In step S407, it is checked whether or not the square correlation bottom position i ₀ ^{S is} present (a square correlation function value that is a minimum value has been detected). When it is determined that the square correlation bottom position i ₀ ^{S is} present (S407: YES), quadratic function interpolation is performed using the three-point square correlation function values H ₋₁ ^S , H ₀ ^S, and H ₊₁ ^S. and calculates the squared real and bottom positions minimum value is obtained squared correlation bottom position i ₀ ^S shift amounts beta ^S of the correlation function value (S409), the image shift amount to further correspond to the true bottom position ( i ₀ ^S + β ^S ) is calculated and the process returns (S411, RET). When it cannot be determined that the square correlation bottom position i ₀ ^S has been detected (S407: NO), linear interpolation is performed using the three correlation function values H ₋₁ , H ₀ and H ₊₁ of the linear correlation function values. To calculate a shift amount β between the bottom position where the true minimum value of the linear correlation function value is obtained and the bottom position i ₀ (S413), and further calculate an image shift amount i ₀ + β corresponding to the true bottom position. Calculate and return (S415, RET).

  According to the automatic focus detection 2 of the second embodiment of the present invention, the linear correlation calculation and the square correlation calculation are simultaneously performed. Therefore, the linear correlation function and the square correlation function are simultaneously obtained by one correlation calculation (shift calculation). Therefore, the calculation time can be shortened.

  FIG. 11 is a flowchart showing a third embodiment (automatic focus detection 3) of the automatic focus detection operation of the present invention. The automatic focus detection 3 is different from the automatic focus detections 1 and 2 of the first and second embodiments in that only the square correlation calculation is performed to detect the square correlation bottom position of the square correlation function value.

  When the automatic focus detection 3 is entered, first, a pair of image signal sequences acquired from the line sensor IH is passed through a low-pass filter (S501), and contrast is calculated at least for the image signal sequence of the reference line sensor A (S503). Subsequently, it is checked whether or not the contrast is OK (S505). If it is determined that the contrast is not OK (S505: NO), the automatic focus detection operation (automatic focus detection 3) is terminated (END).

  When it is determined that the contrast is OK (S505: YES), the pair of image signal sequences is passed through the high-pass filter and the process proceeds to step S509 (S507, S509).

In step S509, a square correlation function value is calculated for the pair of image signal sequences by the image shift method to obtain a square correlation function value (S509). For the obtained square correlation function value, a square correlation function value H ₀ ^S that is a minimum value and a square correlation bottom position i ₀ ^{S from} which the square correlation function value H ₀ ^S is obtained are detected (searched); The detected square correlation bottom position i ₀ ^S and the three-point square correlation function values H ₀ ^S , H ₋₁ ^S and H ₊₁ ^S (see FIG. 6) sandwiching the square correlation bottom position i ₀ ^S are stored. Then, the process proceeds to step S513 (S511, S513). In step S513, the reliability of the square correlation bottom position i ₀ ^S is calculated, and the process proceeds to step S515 (S513, S515). The reliability of the bottom position i ₀ ^S depends on the inclination angle of the straight line passing through the square correlation function values H ₋₁ ^S and H ₀ ^S or the inclination angle of the straight line passing through the correlation function values H ₊₁ ^S and H ₀ ^S , or the square. The correlation function value H ₀ ^S is obtained by a known method such as comparing the magnitude of the correlation function value H ₀ ^S with a threshold value.

In step S515, it is determined whether or not the reliability of the square correlation bottom position i ₀ ^S is OK (present), and when it is determined that the reliability is OK (present) (S515: YES), the three-point square correlation function value H _0. ^A deviation amount β ^S is calculated by a quadratic function interpolation calculation using ^S ₁ , H ₋₁ ^S and H ₊₁ ^S (S517), an image deviation amount i ₀ ^S + β ^S is calculated, and the process returns (S519, RET). . If it is not determined that the reliability is OK (present) (S515: NO), the process skips steps S517 and S519 and returns (RET).

As described above, according to the automatic focus detection 3 process as the third embodiment of the present invention, the image shift amount i ₀ ^S + β is obtained by the square correlation calculation and the quadratic function interpolation calculation without performing the linear correlation calculation. ^{Since S} is calculated, the focus detection process can be simplified.

  As described above, the embodiment in which the automatic focus detection method and the automatic focus detection apparatus of the present invention are applied to a single-lens camera has been described. However, the present invention can be applied to any pupil division phase difference type optical system.

11 camera body 31 body CPU (focus detection calculation means)
32 AF motor driver 33 AF motor 34 Gear block 35 Joint 37 Encoder 38 Flash memory 51 Shooting lens 52 Focus adjustment lens group 53 Gear block 55 Joint 57 Lens CPU
60 AF module 61 Focus detection element 70 Shooting screen CC Focus detection area IH Line sensor A Reference line sensor (reference line sensor area, one area)
B Reference line sensor (reference line sensor area, other area)

Claims (16)

A pair of subject images that have passed through different pupil positions of the photographic lens are projected onto a pair of line sensors having corresponding pixel rows, and the pixel signal values related to the pair of image signal rows by the pair of line sensors are relatively compared. A phase difference method for calculating a correlation amount by shifting, obtaining a phase difference between a pair of image signal sequences based on a shift position where the correlation amount is an extreme value, and obtaining a defocus amount of the photographing lens from the phase difference In the automatic focus detection method,
Obtaining a difference signal value of one and the other of the corresponding pixel signal values of the pair of line sensors;
Performing a square correlation calculation to obtain a sum of squares of the difference signal values to obtain a square correlation function value of the pair of image signal sequences;
Obtaining a square correlation shift position having the highest degree of matching between a pair of image signal sequences from the square correlation function value, and obtaining the phase difference from the square correlation shift position;
An automatic focus detection method characterized by comprising: 2. The automatic focus detection method according to claim 1, wherein the square correlation function is a square correlation function value formed by a sum of squares of difference signal values of one of the pair of image signal strings and the other of the pair of image signal strings. Automatic focus detection that sequentially shifts one of the other from the initial position and obtains a plurality of shift positions and determines that the degree of coincidence of the pair of image signal sequences is higher as the square correlation function value is smaller Method. 3. The automatic focus detection method according to claim 2, wherein the square correlation shift position performs a quadratic function interpolation operation using a plurality of square correlation function values including a minimum value of the square correlation function value, An automatic focus detection method obtained by a shift position at which a true minimum value of a square correlation function value is obtained. The automatic focus detection method according to claim 2, further comprising: before the step of obtaining the square correlation function value, a linear correlation function value comprising a sum of difference signal values of one of the pair of image signal sequences and the other of the pair of image signal sequences. Performing a linear correlation operation obtained by sequentially shifting the other of the image signal sequence with respect to one of the plurality of shift positions, and detecting a minimum value from the linear correlation function values obtained at the plurality of shift positions; And in the step of calculating the square correlation, an automatic focus detection method for obtaining a square correlation function value by performing the square correlation calculation at a predetermined number of shift positions including the shift position at which the minimum value is obtained. . 3. The automatic focus detection method according to claim 2, wherein the step of obtaining the square correlation function value includes calculating a linear correlation function value consisting of a sum of difference signal values of one of the pair of image signal sequences and the other of the pair of image signal sequences. A step of performing a linear correlation operation for sequentially obtaining one of the plurality of shift positions and obtaining a linear correlation function obtained at a plurality of shift positions. After this step, a minimum value is obtained from the linear correlation function values obtained at the plurality of shift positions. A step of detecting, a step of determining whether or not the square correlation function value is a minimum value at a shift position where the minimum value of the linear correlation function value obtained by the linear correlation calculation is obtained, and a minimum value If it is determined, a step of obtaining a square function shift position at which a minimum value of the square correlation function value is obtained, and a case where the minimum value is not determined (when the minimum value of the linear correlation function value is obtained) Schiff Step a, automatic focus detection method comprising the steps of obtaining a square function shift position minimum value is obtained for the square correlation function values to implement the re-squaring the correlation operation on the shift position of a predetermined number including the position. 6. The automatic focus detection method according to claim 4, wherein the predetermined number when the square correlation function value is obtained at a predetermined number of shift positions including the shift position at which the minimum value of the detected linear correlation function value is obtained is: An automatic focus detection method in which the number is larger than the number of shift positions used in the quadratic function interpolation calculation or linear correlation interpolation calculation. 6. The automatic focus detection method according to claim 4, further comprising: after the step of obtaining the square correlation function value, detecting a minimum value of the square correlation function value; and detecting the detected square correlation function value. An automatic focus detection method comprising a step of performing a quadratic function interpolation operation using a predetermined number of square correlation function values including a local minimum value to obtain a shift position at which a true local minimum value of the square correlation function value is obtained. 6. The automatic focus detection method according to claim 4, further comprising a step of detecting a minimum value of the square correlation function value obtained by the square correlation calculation after the step of obtaining the square correlation function value, and a minimum value. And a quadratic function interpolation operation using a predetermined number of square correlation function values including the minimum value of the detected square correlation function value when it is determined that the minimum value has been detected. And determining the shift position at which a true minimum value of the square correlation function value is obtained, and determining that the minimum value of the square correlation function value could not be detected, the minimum value of the linear correlation function value is An automatic focus detection method comprising a step of performing a linear interpolation operation using a predetermined number of linear correlation function values including a shift position for obtaining a true minimum value of the linear correlation function value. A focus detection optical system that projects a pair of subject images that have passed through different pupil positions of the taking lens onto a pair of line sensors having corresponding pixel rows;
A correlation amount is calculated by relatively shifting pixel signal values related to a pair of image signal sequences by the pair of line sensors, and a phase difference between the pair of image signal sequences based on a shift position where the correlation amount becomes an extreme value. A focus detection calculating means for obtaining a defocus amount of the photographing lens from the phase difference;
In a phase difference type automatic focus detection device having
The focus detection calculation means obtains a difference signal value of one and the other of the corresponding pixel signal values of the pair of line sensors, performs a square correlation calculation to obtain a sum of squares of the difference signal values, and performs the pair correlation calculation. A square correlation function value of the subject image signal sequence is obtained, a square correlation shift position having the highest degree of coincidence between the pair of image signal sequences is obtained from the square correlation function value, and the phase difference is calculated from the square correlation shift position. An automatic focus detection device characterized in that 10. The automatic focus detection method device according to claim 9, wherein the square correlation function is a square correlation function value formed by a sum of squares of difference signal values of one of the pair of image signal sequences and the other of the pair of image signals. An automatic focus detection apparatus, which is a function for sequentially shifting one of the columns from the initial position to obtain a plurality of shift positions, and the degree of coincidence between the pair of image signal sequences increases as the square correlation function value decreases. The automatic focus detection apparatus according to claim 10, wherein the square correlation shift position performs a quadratic function interpolation operation using a plurality of square correlation function values including a minimum value of the square correlation function value, An automatic focus detection apparatus which is obtained by a shift position where a true minimum value of a square correlation function value is obtained. 11. The automatic focus detection apparatus according to claim 10, wherein the focus detection calculation means further includes a linear correlation composed of a sum of difference signal values of one of the pair of image signal sequences and the other before obtaining the square correlation function value. Performs linear correlation operation to obtain function values at multiple shift positions by sequentially shifting the function value of one of a pair of image signal sequences, and detects the minimum value from the linear correlation function values obtained at multiple shift positions And in the square correlation calculation, an automatic focus detection apparatus that calculates the square correlation function value by performing the square correlation calculation at a predetermined number of shift positions including the shift position at which the minimum value is obtained. 11. The automatic focus detection apparatus according to claim 10, wherein the focus detection calculation means further obtains the square correlation function value and comprises a linear correlation function consisting of a sum of difference signal values of one and the other of the pair of image signal sequences. A linear correlation calculation is performed at a plurality of shift positions by sequentially shifting one of a pair of image signal sequences with respect to the other, and then a minimum value is obtained from the linear correlation function values obtained at a plurality of shift positions. When the square correlation function value is the minimum value at the shift position where the minimum value of the linear correlation function value obtained by the linear correlation calculation is obtained, the minimum value of the square correlation function value is obtained. If the square function shift position is obtained and the minimum value is not the minimum value, the square correlation calculation is performed again for a predetermined number of shift positions including the shift position at which the minimum value of the linear correlation function value is obtained. Function value Automatic focus detection device for determining the square function shift position where the small value is obtained. The automatic focus detection apparatus according to claim 12 or 13, wherein the predetermined number when the square correlation function value is obtained at a predetermined number of shift positions including the shift position at which the detected minimum value of the linear correlation function value is obtained is: An automatic focus detection apparatus having a larger number than a plurality of shift positions used for the quadratic function interpolation calculation or linear correlation interpolation calculation. 14. The automatic focus detection apparatus according to claim 12, wherein the focus detection calculation means further detects a minimum value of the square correlation function value after obtaining the square correlation function value, and detects the detected square. Automatic focus detection apparatus for performing a quadratic function interpolation operation using a predetermined number of square correlation function values including the minimum value of the correlation function value to obtain a shift position at which a true minimum value of the square correlation function value is obtained. . 14. The automatic focus detection apparatus according to claim 12, wherein the focus detection calculation means further detects a minimum value of the square correlation function value obtained by the square correlation calculation after obtaining the square correlation function value. If it is possible, a true minimum value of the square correlation function value is obtained by performing a quadratic function interpolation using a predetermined number of square correlation function values including the detected minimum value of the square correlation function value. When the shift position is obtained and the minimum value of the square correlation function value cannot be detected, linear interpolation is performed using a predetermined number of linear correlation function values including the minimum value of the linear correlation function value, and linear correlation is performed. An automatic focus detection device for obtaining a shift position at which a true minimum value of a function value can be obtained.

Images