JPH0553241B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention is an automatic focus detection device used in optical systems such as microscopes and cameras. The present invention relates to an automatic focus detection device that performs focus detection by individually converting into photoelectric output signals using a conversion device and detecting the relative positional relationship between two images based on the photoelectric output signals.

Conventional automatic focus detection devices of this type include a distance type that applies triangulation and a TTL type that divides the light flux passing through the pupil to obtain two images, and both detect the coincidence of the two images. In order to do this, the correlation between the two images was calculated digitally, the peak of the correlation value was taken as a match, and the amount of relative movement of the two images until the match was taken as the amount of phase difference between the images. .

Figure 1 shows an example of this, in which data A and B of two images taken by an image sensor (not shown) are passed through a sample hold circuit, an A-D converter (none of which is shown), etc., into a ring shape. are stored in shift registers 1a and 1b, respectively. In this example, the image data consists of 128 bits. Both image data A and B are stored in shift registers 1a and 1, respectively.
Once stored in bit b, the circuit 2 for calculating the absolute value of the difference calculates the absolute value of the difference for each bit.
Further, an adder 3 calculates the sum of the two images to obtain a correlation value between the two images. Next, the image data B in the shift register 1b is shifted by one bit by a pulse from the clock CL, and the circuit 2 and the adder 3 are moved again.
The correlation value is obtained by In this way, each time one of the image data is moved one after another by the clock CL, a correlation value is determined, and the peak detector 4 further determines the extreme value of the correlation value, and the position where the extreme value is detected is the same. It becomes the focal position. Further, the number of clocks in the case of an extreme value is determined by the counter 5,
This number of clocks, that is, the amount of movement of the image data B in the shift register 1b becomes the amount of phase difference between the two images.
From this, you can know the direction and amount of focus.

However, in this conventional device, since the image sensor has to have a certain size, the two images captured on the image sensor are not only shifted in position, but also have different peripheral parts, and as a result, the two images are different. The image data A and B stored in the two shift registers 1a and 1b are not only shifted in position, but also different at the ends, and the correlation is calculated while rotating one of the image data in a ring shape. Therefore, it was not possible to obtain an accurate image phase difference. To explain this point in detail based on FIG. 2, a and b indicate image data A stored in the shift register 1a and image data B stored in the shift register 1b, respectively. If not, both image data A and B are shifted, and therefore the peaks P and P' are also shifted, and both image data A and B are different at the ends. Further, c indicates the image data B in the shift register 1b shifted by α bits. Therefore, since the peaks P and P'' match, it can be seen that the phase difference between both image data A and B is α bit, but among the image data of c, 0 to
The .alpha.-bit portion is the .beta. to 127-bit portion in state b, and the image data of a and c do not completely match. That is, when the correlation is calculated for all image data of 0 to 127 bits, the correlation value does not always reach an extreme value when the image phase difference is zero, that is, when the peaks P and P'' match. Therefore, it is difficult to obtain an accurate image phase difference with this device.Also, with this device, the movement of image data is
Since this is done only bit by bit, it also has the disadvantage that it takes time to detect a large amount of defocus.

In view of the above-mentioned problems, the present invention detects the relative positional relationship using photoelectric output signals of some elements of a photoelectric conversion device, and when the amount of defocus is large, a plurality of elements of the photoelectric conversion device are shifted by a plurality of elements. The purpose of this invention is to provide an automatic focus detection device that detects the relative positional relationship using the photoelectric output signal of This will be explained based on an embodiment shown in Fig. 15. Fig. 3 shows the principle of the pupil division method used in this embodiment, and in a, 6 is an imaging lens, and 7 is an imaging lens. 6 is a light shielding plate having an aperture 7a disposed near the pupil, 8 is an image plane, and when in focus, an image Q is formed on the image plane 8.
However, when out of focus, blurred images Q ₁ and Q ₂ are shifted in opposite directions perpendicular to the optical axis 0 with respect to the front focus and rear focus, respectively, at the image plane 8.
formed on top. b shows the case where the aperture 7a of the light shielding plate 7 is moved to the opposite side with respect to the optical axis 0, and when in focus, an image Q' is formed on the image plane 8;
When out of focus, blurred images Q ₁ ' and Q ₂ ' are formed on the image plane 8 corresponding to the front focus and the rear focus, respectively. Therefore, the opening 7a of the light shielding plate 7 can be moved, for example, from the position a to the position b.
If you move it to the position, the image Q and
Q′ stays at the same position and does not move, but if the front focus is on the image, the image moves from Q ₁ to the Q ₁ ′ position, and if the back focus is on the image, it moves from Q ₂ to the Q ₂ ′ position. . By providing a so-called image sensor on the image plane 8, the state of the image can be measured.

From the above, it is possible to distinguish between front focus and rear focus, and to know the amount of out-of-focus at that time from the moving direction of the image and its amount (ie, phase difference).

FIG. 4 is an overall view of an embodiment in which the above principle is applied to a microscope, where 51 is a light source and 52 is a condenser lens. 53 is a stage on which the specimen is placed for observation. 54 is an objective lens;
55 is a beam splitter that guides light to the focus detection system, 56 is a prism that guides a portion of the light to an eyepiece, and 57 is an eyepiece. 58 is a photographic eyepiece lens, and 59 is a film. Up to this point, the optical system is no different from an ordinary microscope. 60 is a relay lens that guides the light from the beam splitter 55 to the detection optical system, 61 is a lens for forming an image of the pupil of the objective lens 54 at a predetermined position together with the relay lens 60, and 62 is the position of the pupil image. It is a pupil divider placed in the pupil divider. Reference numeral 63 denotes an imaging lens, and the light passing through this lens forms an image on an image sensor 65 via a filter 64. 66 is a pupil divider drive circuit, and 67 is a stage drive circuit, each of which is controlled by a microcomputer 70. 68 is an image sensor drive circuit, and 69 is an interface circuit that inputs image data from the image sensor 65 to the microcomputer 70. 71 performs automatic focus operation,
This is a console that displays in-focus and impossible indicators. The microcomputer 70 performs all correlation calculations, focus determination, etc. Correlation calculations can be performed using calculation-dedicated LSIs that have recently been developed and are on the market.

Next, the operation of each part will be explained in detail. Figure 5 shows two specific examples of a pupil splitter that splits the light flux passing through the pupil to form two images, and a shows a pupil splitter in which a light-shielding part (shaded part) is provided on the glass substrate and the axis is 0.
The pupil 9 is opened and closed alternately in half by rotating the pupil 9 around the axis 0, and b is a sector-shaped one with an opening 10, and by rotating left and right around the axis 0, the pupil 9 can be opened and closed in half. They are designed to open and close alternately. Case a is suitable for a method in which an image sensor is used to capture an image in response to a synchronization signal from a pupil splitter that is rotated by a DC motor or the like. b is suitable for a method in which the pupil divider is moved according to a control device such as a microcomputer and an image is captured by an image sensor in synchronization with the movement of the pupil divider. In this way, the states a and b in FIG. 3 are created using the pupil divider as described above, and the states a,
It becomes possible to obtain image data for each of the states b using the image sensor.

Furthermore, since the specimen to be focused on is generally not necessarily located at the center of the field of view, it is desirable that the image sensor cover as wide a range as possible and not just at the center of the field of view. However, in order to cover the entire field of view, it is necessary to increase the number of elements in the image sensor. This comes from the fact that the pitch of the elements must be set to a certain size in order to maintain constant focusing accuracy.

This point will be explained below. FIG. 6 is a diagram showing the relationship between the amount of phase difference between two images and the amount of defocus. Here, to simplify the explanation, we will consider a point image. 13 is the optical axis of the optical system, and it is assumed that the point image 11 is formed by the optical system whose rear numerical aperture is NA'. Assuming that the image sensor 12 is now located at the position of the defocus amount δd, two images 11
Since A and 11B are formed with a phase difference of Sp, the relationship between δd and Sp is δd=Sp/NA' (1). Now let's consider the focusing accuracy when using a 10x objective lens. If the NA of the 10× objective lens is 0.40, the NA′ is 0.04, which is derived from equation (1) as δd=25Sp (2). On the other hand, the depth of focus t of the 10× objective lens
is expressed as t=ε/NA′ (3) (ε is the permissible circle of confusion), so ε
= 0.05mm (equivalent to a resolution of 20 lines/mm), then t = 1.25mm (4). Since focus detection accuracy within this depth of focus is required, δd = 1/2 (5), δd = 0.625 mm (6), and therefore Sp = 25 μm (7). In order to accurately obtain this amount of image phase difference, the pitch of the diode array of the image sensor 12 must be approximately 25 μm.

As described above, the pitch of the image sensor 12 is determined based on the requirement for focusing accuracy. In this case, temporarily 128
If we use an image sensor with a diode array of
It becomes necessary to move the subject to be focused on to the position (generally the center) of the image sensor to perform focusing.

FIG. 7 shows the image sensor used in this example and the processing method used therein. This embodiment uses an image sensor having 512 photodiode arrays, which has a size of 512×0.025=12.8 mm, which can cover a considerable portion of the field of view. However, all bits (diode array)
If the correlation calculation is performed using , the calculation time will be extremely long and it will be meaningless. Therefore, the 512 bits are divided into five blocks a to d of 128 bits, and the correlation calculation is performed on the block with the highest contrast.

Here, an example of a contrast calculation method will be described. Generally, as an evaluation function for contrast evaluation, when the x-th bit output of the f(x) sensor is used, C= 〓 ^x | f(x) - f(x+1) | or C= 〓 ^x | (f(x) -f(x+1)) ² is known, and in this example, unlike focusing using the contrast method, which requires accurate knowledge of changes in contrast, it is only necessary to know the relative contrast strength between each block. Therefore, it is not necessary to calculate the difference between adjacent bits. For example, if C= 〓 ^x ′|f( ^x )−f(x+ ₅ )| ^x=64 |f(x)-f(x+5)| =|f(64)-f(69)|+... +|f(184)-f(188)|, and the absolute value of the difference is calculated 31 times. All you have to do is add it up while calculating. With the conventional calculation method, 121 times would be required.

Note that the reason why the absolute value of the difference between 5-bit adjacent values is calculated every 4 bits is to improve the contrast sensitivity, rather than simply calculating the absolute value of the difference between adjacent values every 4 bits. It's to make things better. Regarding this point, if we perform a comparative calculation based on the light intensity distribution l on the image sensor shown in FIG. 8, for example, in the case of this example, C=|f(64)-f(69)|+ ｜f(68)−f(73)
| = | 13 − 30 | + | 25 − 60 | = 52, whereas in the conventional case C′ = | f (64) − f (65) | + | f (68) − f (69
) | = | 13 - 14 | + | 25 - 30 | = 6 Therefore, the contrast sensitivity of this embodiment is better than that of the conventional case.

Further, in order to reduce the amount of calculation, when calculating the difference between X bits and the adjacent value and calculating this every Y bits, it is preferable that X>Y. In this example, X=5>
4=Y.

If the above is expressed as a general formula, C= _〓o |f(xn+z)−f(xn+z+y)|. That is, if n = 0, 1, 2,..., the position of the element changes every x, and by looking at the output difference between each element at every x and the element that is separated by y. , the contrast will be evaluated. However, x and y are positive integers, z is zero or a positive integer, and x>y, x≠z.

Note that the data used for the function f may be either image data A or B.

Thus, blocks a, b,
..., e and select the one with the best contrast, but as it is clear from Figure 7, blocks a and b overlap with 128 to 192 bits, so it is wasted. Calculate the contrast of 64 to 128 bits, 128 to 192 bits, and 192 to 256 bits to avoid calculations.The contrast of block a is the sum of the contrast of 64 to 128 bits and 128 to 192 bits, and the contrast of block b is the sum of the contrast of 64 to 128 bits and 128 to 192 bits. It may also be the sum of the contrasts of 128 to 192 bits and 192 to 256 bits. Furthermore, blocks a, b, c,
The reason why half of d and e overlap each other is to enable setting of a block that includes a significant change in image intensity even if there is a significant change in image intensity at the boundary of the block. For example, if there is a significant change in image intensity near the boundary between blocks a and b, i.e. around 192 bits, all of the information cannot be used in block a or block c, but if block b is set, all of the information will be used. is included in block b, which is convenient. Since the computation for determining the contrast takes a much shorter time than the computation for the correlation, the computation time in this embodiment is about 128-bit correlation computation time + α. or,
The reason why blocks are not set for each of the 64 bits on both sides is to prevent the situation described in FIG. 2 from occurring when the correlation is calculated by shifting the images.

By performing the processing as described above, even if the subject or specimen to be focused is not necessarily in the center of the field of view, a certain part (block) of the subject is automatically selected and focus detection is performed. For the above, there is no need to define a fixed block, and from among the many photodiode arrays of the image sensor arranged to cover most of the field of view,
The portion containing the photodiode array required for the correlation calculation may be selected based on contrast or the like. Alternatively, a mark or the like may be provided in the field of view for observation and the setting may be made manually. In this way, even if there is a three-dimensional specimen or dust in the field of view, the user can focus on the subject he or she wants to focus on.

Next, the overall operation will be explained. First, two image data A and B from the image sensor 65 in FIG. 4 are stored in the memory of the microcomputer 70 through the interface 69.
Then, the block with the highest contrast among the five blocks is selected, and the correlation is calculated using the image data of that block. Let us proceed by assuming that block a in FIG. 7 is selected.

In the correlation calculation, image data A and image data B corresponding to two images stored in memory are calculated by relatively shifting them by 1 bit, and it is determined how many bits have to be shifted before the images overlap. Find the amount of phase difference. The correlation formula is, for example, R(δ)= ₁₉₁ 〓 ^x=64 ABS{f _A (x)−f _B (X+δ)} (8) where ABS represents the absolute value and the functions f _A (x), f _B (x)
represent the x-th bit values of image data A and B, respectively. And for a set of functions f _A , f _B δ
The phase difference is defined as δ, that is, δ', when R(δ) is minimized by changing . Also, in this example, −64≦δ
≦64. Since the range of δ can be narrowed near the focal point, calculation time can be shortened.

Since the actual value of δ is only taken for each bit of the image sensor, in order to detect even more accurately, the discrete value of the correlation is approximated by curve fitting, etc. to obtain an image with an accuracy of 1 bit or less. Find the phase difference (Figure 9). Alternatively, it can also be determined by performing quadratic curve approximation using δ when R(δ) is minimum, that is, δ', and three points o, p, and q before and after it. (Figure 10).

As a result of the above, it is possible to cover a large part of the field of view and maintain focusing accuracy while hardly increasing the calculation time.

In the above example, δ is in the range of -64 to 64, and the defocus amount in this range is 0.625 × 64 = 40 from equation (6).
mm, and since 40/10 ² =0.4=400 μm on the objective side, the amount of defocus is ±400 μm. If it is desired to include a wider range of defocus amounts within the detection range, it may be possible to increase the range of δ, but this increases the amount of calculation and is preferable. Furthermore, when the amount of defocus is large, there is no point in performing highly accurate calculations as described above.

Therefore, in this embodiment, contrast and correlation calculations are performed using data for every few bits of the data captured as image data. Specifically, if data is used every 5 bits to make 8, then the data in the image memory is f(0), f(1), f(2),..., f
(510), f(0), f(5), f(10) in f(511),
...,
Just think of f(505) and f(510) as the data to be used. In reality, only every fifth bit of data is used during calculation. For example, equation (8) becomes R(δ)= ₆₄ 〓 ⁿ⁼⁰ |f _A (96+4n)−f _B (96+4n+δ)| (9). In this case, since the amount of defocus is large and the image is blurred and has only low frequency components, it is not divided into blocks, but it may be divided into blocks if necessary. In this case, if the amount of change in δ, that is, the amount of shift between two image data when performing one correlation calculation and then the next correlation calculation, is also calculated using 5 bits each, the range of δ can be set to, for example, −200≦
Even when δ≦200, the number of correlation calculations is 81, which is small. The detection range is ±1.25mm.

As described above, by using image data every few bits and performing the correlation calculation while shifting the bits by several bits, the detection range can be expanded without increasing the amount of calculation. In this way, when the amount of defocus is large, it is calculated every few bits to bring it close to the focal point, and then the calculation taking into consideration the accuracy mentioned above, i.e. 1
By calculating the correlation for each bit and performing curve fitting approximation or quadratic curve approximation, automatic focus detection with a wide focusable range and high focusing accuracy can be performed.

Furthermore, if the amount of defocus is large, it is better to move the stage while calculating only the contrast (which requires less calculation) from the image data, and then perform focusing by correlation after the contrast reaches a constant value. . Particularly in microscopes, compared to cameras and the like, a slight deviation in the position of an object is magnified on the image plane, so that a severe defocus state often occurs. For this reason, it is recommended to perform coarse focusing to the extent that correlation calculation is possible using the contrast detection method, which requires a short calculation time, and then quickly use the phase difference method, which can calculate the amount of movement to the in-focus position with a single calculation. This is effective for accurate focusing. Furthermore, when the contrast of an image is detected, if the contrast is not above a certain level, there is a risk that the correlation calculation will calculate an incorrect image phase, so it may be used to determine not to calculate if the contrast is lower than that.

In the case of a device such as a camera that moves an objective optical system to perform focusing, it goes without saying that the optical system is driven.

In the case of the above embodiment, since the light flux passing through the pupil is divided to obtain two images, the light amounts of image data A and B may differ due to eccentricity of the optical system, eccentricity of the pupil, etc.
This is especially likely to be the case if the focusing system is attached. Furthermore, if there is no pupil divider at the pupil position, the amount of light will be uneven between image data A and B as shown in FIG. FIG. 12 is a diagram schematically illustrating unevenness in light amount. In the case of a, since the pupil and the pupil divider coincide, the amount of light passing through the pupil for each image height h, i, and j is all equal to a. In case b, since the pupil and the pupil divider do not match, the amount of light passing through the pupil for each image height h, i, and j becomes b, a, and c, respectively, and becomes nonuniform, resulting in uneven light amount. .

If there is a difference in light amount or unevenness in light amount as described above, the similarity between the two images of image data A and B deteriorates, and the accuracy of the correlation processing result is significantly reduced. Therefore, correction is necessary. One example of correction is a method commonly used to remove fixed pattern noise from image sensors, in which image data is obtained by exposing an image sensor to uniform light in advance.
Since the image data is the fixed pattern noise itself due to the uniformity of the incident light, if a correction coefficient is created by the reciprocal of the image data, then the influence of the fixed pattern noise can be removed by multiplying the correction coefficient by the image data. . In the case of this embodiment, image data A and B are created using uniform light passing through the focusing optical system.
When obtained, image data A and B become 11th due to eccentricity etc.
As shown in the figure, the data has uneven light intensity. Therefore, by creating a correction coefficient using its reciprocal or the like and performing similar processing, the influence of uneven light amount can be removed. Even if there is a difference in light intensity, the same effect can be obtained by performing the same processing through the optical system. Incidentally, it also serves to remove fixed pattern noise from the image sensor. As a specific method for exposing with uniform light, a simple method is to input image data with no sample placed on the stage 53.

As described above, the above correction requires data input using uniform light once. This is not a troublesome operation, but if you still do not want to input data for correction, you can also correct it by calculation. Figure 13 is a diagram explaining this.
The x-axis is aligned in the direction of arrangement of the sensor array, and the y-axis is aligned in the direction of the intensity of image data. As can be seen from the explanation of FIG. 12, the values of image data A and B can be thought of as straight lines with a certain slope.
Let them be l _A and l _B , respectively. When the slope of the image data l _A is α _A , the formula for the image data l _A is y=β _A x+I _A , where I _A is the average of the light amount of l _A. Although the slope β _A changes depending on the amount of light, it can be determined from k by introducing a constant k such that β _A =I _A /K. k is determined by the characteristics of the optical system, and can be measured in advance. The average light amount of image data A and B is I _a ll,
The correction coefficient α is α=I _a ll/I _A /Kx+I _A =I _a ll/I _A ·1/x/k+1, thereby making it possible to correct the light amount difference and the light amount unevenness.

As described above, by using uniform light or performing calculations, it is possible to correct and eliminate the effects of eccentricity of the optical system or eccentricity of the pupil, or the effects of misalignment between the pupil and the pupil divider. As a result, focusing accuracy is improved and the detection range is expanded. Furthermore, the focus detection section can also be used in the form of an attachment. The most significant effect is that objective lenses of various magnifications with different pupil positions can be used.

Furthermore, when many types of objective lenses are used, such as in a microscope, the pupil positions differ depending on each objective lens, so it is difficult to set the pupil divider correctly for all the objective lenses at the pupil positions. One possible solution to this problem is to provide a plurality of pupil dividers. By providing a plurality of pupil dividers at the pupil position of each objective lens, the pupils and the pupil dividers are made to coincide. It goes without saying that when one pupil divider is used, the other pupil dividers are configured so as not to disturb the pupil. . For example, the first
As shown in FIG. 4, the same two pupil dividers as shown in FIG. 5 connected together may be used.

In FIG. 4, a filter 64 is an infrared cut filter or a band pass filter, and serves to prevent the phenomenon of defocus caused by the spectral sensitivity and spectral distribution of the image sensor, light source 51, etc., which differ from the relative sensitivity. do.

When controlling and arithmetic processing of the automatic focusing device as described above, a method using a microcomputer and arithmetic processing unit is easiest and cheapest in design.
A supplementary explanation of this will be given with the flowchart shown in FIG. This shows the most basic case. When focusing is started, it is first checked to see if the microscope is in a state suitable for focusing, and the type and magnification of the objective are determined. This is because in the case of light intensity unevenness correction, the parameters differ depending on the type of objective and the magnification, and the conversion coefficient for converting the amount of image phase difference into the amount of stage movement also differs depending on the magnification (see equation (1)). . Next, from the image sensor f _A
and f _B data is obtained and stored in memory. Thereafter, the unevenness in the amount of light is corrected and the image is stored in the memory again. Since the focus may be largely out of focus at the start of focusing, the approximate focus position is determined by correlation calculation every 5 bits (see equation (9)). Then, the image phase difference amount determined by the correlation is converted into a stage movement amount, and the stage is moved. Therefore, data on f _A and f _B is obtained again and correction is performed. Next, blocks are determined by contrast evaluation. If the contrast is not above a certain value, the reliability of the correlation results will be low.
Perform 5-bit correlation again and move the stage closer to the focal position. If the contrast does not increase even after several attempts, the contrast of the sample is too low, and an "impossible" indication is displayed. If the contrast is above a certain level, the correlation is calculated using the determined block, and the stage is moved to focus. To confirm focus, obtain f _A and f _B again and calculate the correlation. Here, if the image phase difference amount is within the depth of focus, the image is in focus and the stage is not moved. If it is outside the depth of focus, repeat the same operation again.

The above is an explanation of the most basic operation, and the actual program also takes into consideration phenyl safety in the case of no sample or machine failure.

Further, the amount of image phase difference may be used for determination when changing from 5-bit correlation to 1-bit correlation. In the previous example, the range of -200≦δ≦200 is calculated every 5 times, but δ where the correlation R(δ) takes the minimum value is -200≦δ′≦
If it is 200, move the stage by that amount and then move to 1-bit correlation. In this case, the judgment condition is y180≦δ′≦80
It is better to set it to a value smaller than the range of δ calculated as follows. This is because when the amount of defocus is large, δ' may be incorrectly determined to take the minimum value due to noise or the like.

FIG. 16 shows a second embodiment of a control/arithmetic circuit in which components other than the central processing unit are constructed of hardware. To explain this, first, the pupil divider drive circuit 66 operates in response to a focus start signal from the console 71 and image data A is obtained by the image sensor 65. The image sensor 65 starts imaging by the sensor drive circuit 68 in synchronization with the pupil splitter 62. At this time, in the case of a storage type image sensor (solid-state image sensors are generally of this type), color readout is performed to erase previously stored signals. Image data A continuously read out from the image sensor 65 is stored in the first memory 34 through the sample hold circuit 31, A/D converter 32, and switch circuit 33. Then, the image data A is corrected using correction coefficient number data stored in advance in a memory (not shown) and stored in the first memory 34 again. The correction coefficient data is obtained by multiplying the reciprocal of the above-mentioned image data captured by the image sensor with uniform light incident thereon by the average value of the image data A and B at that time. Consider the case of a 512-bit image center. Image data A and B together are 0~
Total of 1024 pieces up to 1023 bits. Image data A
is 0 to 511 bits, and image data B is 512 to 1023 bits. If the value of the n-th bit of image data obtained with uniform light without a sample is x _o , then the correction coefficient k _o of the n-th bit is becomes.

Next, when the image data A is stored in the first memory 34, the pupil divider 62 becomes in a state to take the image data B, and the image data B is corrected and stored in the second memory 35 in the same manner as the image data A. be done. The data stored in the first memory 34 is sent to the contrast discriminator 36 for each block shown in FIG.
The block to be used is determined by the height of contrast. When the contrast of block b is the highest, 128 is given to the addressing circuit 37. The address shift circuit 38 has an initial value of -3.
2, and when the address designation circuit 37 designates 128 of the first memory 34, the address shift circuit 38 designates 96 of the second memory 35. Then, the image data f _A (128) and f _B (96) are transferred to the subtracter 39
|f _A (128)−
The calculation f _B (96) | is performed. And adder 4
1 and then stored in the memory 42. When the above steps are completed, the addressing circuit 37
The address shift circuit 38 specifies the memory 35
97 is specified, and thereafter it is calculated in the same way |f _A −
(129)−f _B (97)| is added to the previous data |f _A (128)−f _B (96)| by the adder 41 and stored in the memory 42.
is stored in From now on, address 2 of the first memory 34
55, and the correlation calculation R(-32)= ₂₅₅ 〓 ^x=128 ABS{f _A (x)-f _B (X-32)} is completed. When this calculation is completed, the value of the address shift circuit 38 becomes -31, and the following is calculated: R(-31)= ₂₅₅ 〓 ^x=128 ABS{f _A (x)-f _B (X-31)}. This continues until the value of the address shift circuit 38 becomes 31, and the entire correlation calculation R(δ)= ₂₅₅ 〓 ^x=128 ABS{f _A (x)−f _B (X−δ)}−32≦ δ≦31. Next, the central processing unit 43 compares R(δ) in the memory 42, finds δ that minimizes R(δ), and uses it as the amount of phase difference of the image. Accordingly, the stage drive circuit 67 is driven to adjust the focus.

In addition, if the amount of defocus is large and the phase difference amount of -32≦δ≦31 is insufficient, the addressing circuit 37
specifies 128 in the first memory 34, and the initial value of the address shift circuit 38 is -64. and,
It increases by 2 after each correlation calculation, R (-64) = | f _A (128) - f _B (128-64) | + | f _A (130) - f _B (130-64) | + ...+|f _A (382)-f _B (382-64)| +R(-62)=... 〓 R(60)=... The following calculation is performed. This converts the image data into 1
This corresponds to using every other bit, and the detection range of the image phase difference is doubled with the same amount of calculation. However, the focusing accuracy will be 1/2.

As described above, the image phase difference is calculated and the stage drive circuit 67 is driven to adjust the focus, but the above operation may be repeated several times in order to accurately adjust the focus. Note that the console 71 starts focusing and displays the focus.

As described above, the automatic focus detection device according to the present invention performs a correlation calculation using only a part of the image data obtained by the image sensor, and when the amount of defocus is large, it uses data shifted by multiple pieces of image data. Since the correlation calculation is performed based on the data, the detection accuracy and detection speed are greatly improved.

It goes without saying that the present invention can also be applied to an automatic focus detection device using two image sensors.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the main circuit of a conventional automatic focus detection device, FIG. 2 is a diagram showing an example of image data stored in the shift register of the conventional example, and FIG. 3 is an automatic focus detection according to the present invention. 6 is a diagram showing the principle of the pupil division method used in one embodiment of the device; FIG. 4 is an overall view of the above embodiment; FIG. 5 is a front view showing a specific example of the pupil splitter used in the above embodiment; The figure shows the relationship between the amount of phase difference between two images and the amount of defocus.
Fig. 7 is a diagram showing the image sensor of the above embodiment and the processing method therein, Fig. 8 is a diagram showing an example of the light intensity distribution on the image sensor, and Figs. 9 and 10 are accuracy near the focal point. Fig. 11 is a diagram showing the unevenness of light amount of image data in the above embodiment, Fig. 12 is a diagram schematically explaining the unevenness of light amount, and Fig. 13 is a method of correcting the unevenness of light amount. FIG. 14 is a perspective view of another pupil divider, FIG. 15 is a flowchart showing the computer control and arithmetic processing method of the above embodiment, and FIG. 16 is a control/arithmetic circuit of another embodiment. FIG. 51...Light source, 52...Condenser lens,
53...stage, 54...objective lens, 55...
...beam splitter, 56...prism, 57
...Eyepiece, 58...Photography eyepiece, 5
9...Film, 60...Relay lens, 61...
... Lens, 62 ... Pupil divider, 63 ... Imaging lens, 64 ... Filter, 65 ... Image sensor, 66 ... Pupil divider drive circuit, 67 ... Stage drive circuit, 68 ... Image sensor drive Circuit, 69...Interface circuit, 70...Microcomputer, 71...Console.

Claims (1)

[Scope of Claims] 1. Two images, first and second, formed through different optical paths are converted into a photoelectric output signal by a photoelectric conversion device formed by arranging a large number of elements, and based on the photoelectric output signal, In an automatic focus detection device that performs focus detection by detecting the relative positional relationship between two images, the large number of elements is divided into a plurality of blocks each including at least two or more elements, and at least one Regarding the image, the relative positional relationship between the two images is detected by using the photoelectric output signals of the elements at both ends of the photoelectric conversion device or the elements other than the element group, and the photoelectric output representing the first image is detected. a first step of detecting the relative positional relationship between the two images and performing approximate focus detection by comparing the signal and the photoelectric output signal representing the second image by shifting them by a plurality of elements; After the step, the contrast of the first image is calculated based on the photoelectric output signal representing the first image, and the contrast of the first image is calculated from one of the plurality of blocks according to a predetermined contrast. a second step of selecting a photoelectric output signal of one block selected in the second step of the photoelectric output signals representative of the first image after the second step; An automatic focus detection device comprising: a third step for detecting the relative positional relationship between the two images by shifting the photoelectric output signal representing the second image one element at a time and comparing the two images. 2. The automatic focus detection apparatus according to claim 1, wherein the first step is carried out using a photoelectric output signal from at least every plurality of elements selected from among the photoelectric output signals. 3. The automatic focus detection device according to claim 1, wherein an evaluation function shown below is used to detect the contrast of the first image. C= 〓 ⁿ | f (xn + z) - f (xn + z + y) | x>y, x≠z x, y: positive integer, z: zero or positive integer n: 0, 1, 2,...