JPS6313009A - Focus detecting device - Google Patents

Abstract

PURPOSE:To prevent wrong focus detection by judging a focusing state when a correlative value F(S) indicating a deviation quantity S with the largest correlativity and a correlative value F(S-1) or F(S+1) have a desired relation. CONSTITUTION:An object light image passed through image forming lenses L1 and L2 are formed on the light receiving faces of the 1st and the 2nd photoelectric converting element groups 1A and 1B which form picture element arrays A and B of a CCD 1 and respective picture elements begin to be integrated at the same time. When the integral value of a light receiving element reaches a specific value, integral charges of respective picture elements are transferred, AD-converted 3 in order, and stored in memory addresses in a CPU 4 which are assigned to the respective picture elements. Then, a correlation function F(S) is found. The F(Sm) with the smallest value is found therein to find contrast functions C1 and C2. When a relation of correlation value F(Sm+1)>=F(Sm-1) holds, a specific relation is born among those values on condition that the F(Sm) is correct and its value corresponds to the distance between a lens and a body. When this relation is not satisfied, a nonfocusing state is displayed on a display device 7 and the lens is driven by a lens driving control circuit to make a restart.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a focus detection device, and more particularly to a focus detection device that receives a subject light beam and detects a focus state in an optical device such as a camera.

[Prior art] A first part and a second part of a photographic lens sandwiching the optical axis of the photographic lens.
A photographic lens exit pupil split type focus detection device that detects the relative position of two images created by the object light beams that have passed through the respective sections and determines the focus state has already been disclosed in Japanese Patent Application Laid-Open No. 59-1.
It is well known from Japanese Patent Application Laid-open No. 26517 and Japanese Patent Application Laid-Open No. 60-247210. FIG. 12 shows the basic configuration of the optical system.

The pupil projection lens 101 is configured so that the pupil position of the pupil dividing lenses 102 and 103 is conjugate with the pupil position of the photographic lens 100. The pupil splitting lenses 102 and 103 are lenses for re-forming images on the light receiving surface of the second pixel row consisting of the CCD line sensor 105, which is the same as the light receiving surface of the first pixel row consisting of the CCD line sensor 104. It is.

In an optical system configured in this way, the distance between the two images becomes narrower when the front focus is on, and widened when the rear focus is used. At the focused position, the image interval is constant regardless of the photographic lens.

The focus detection device disclosed in the above-mentioned Japanese Patent Application Laid-Open No. 59-126517 has a first pixel column and a second pixel column each composed of a CCD line sensor, and receives an image pattern signal from one pixel column. The image pattern signal is divided into a plurality of image pattern signals each having a specific number of pixels smaller than the number of pixels in the pixel row, and these divided image pattern signals are compared with the image pattern signal from the other pixel row. Divide the first image pattern signal into three block signals, compare each block signal with the second image pattern signal, and calculate the sum of the absolute values of the differences in pixel output, Hi(,,1a)=ΣI''K+si -RK+4-+1
-1 Ωmm1.2,3. -------21 i-0,1,
2, find the minimum value Hi (fIi)0, 1.2.

On the other hand, -1 i-0, 1, 2 is obtained from the contrast function ΣI''K+8i''K+1+8i1, and focus detection is performed based on the minimum value of 1i-0, 1, 2, where 甜i = Hi (/?71). This provides an evaluation value that is not affected by contrast.

Further, the focus detection device disclosed in the above-mentioned Japanese Patent Application Laid-Open No. 60-247210 also uses the minimum value of the correlation value obtained in substantially the same manner as above, and the constant a2. a3.
It compares a2C2, a3C2゜a4C2 multiplied by a4, and determines whether focus detection is possible according to the comparison result to prevent incorrect focus detection when the contrast is low. be.

[Problem to be solved by the invention] By the way, in lenses with a long focal length or macro lenses, which have a large amount of lens extension, it is difficult to detect the defocus state in the entire extension range because the amount of lens extension is large. If this is not possible, there is a risk that a focus detection signal will be output erroneously. In the conventional example described above, the absolute value of the correlation value or the correlation value normalized by the contrast function is unknown, or even if that value is set, it may be insufficient depending on the subject.

Figures 8 and 9 explain the above-mentioned problem in a little more detail.
In the pixel columns of column A and column B, which are composed of CD line sensors,
This is the result of calculating the engine number F(S) by a known method while projecting an optical image with a phase difference of a predetermined pitch and shifting either the A column or the B column by 1 pitch. In Figure 8.9, column A.

The output of each pixel in column B is sequentially output, and the output of column A is al.

a2 ' a3 '"""' a84, bl the output of column B.

b2. ba, ..., b64.

When the shift amount S is S≧0, at in column A.

a21..., a20 is the reference pixel column, B
The same number of columns is shifted by one pitch relative to column A, and the sum of the absolute values of the differences in the outputs of corresponding pixels is calculated.

That is, when S≧0, Hl(S)=ke1an-bn+,1 n- However, in S-0, 1, 2, ..., 32S, the B column is the reference pixel column. Shift column A. That is, in SkuO, H2(S)=Σ 1bo-an+51 S--1, -2, . . . , -32 The contrast function is obtained.

It is.

Shift fl' S ”” S that gives the minimum value of F(S)
o is the amount of lateral shift between the two images. However, since S is an integer value, the precision becomes poor if only this value is used, so it is necessary to perform interpolation. This interpolation method was previously proposed by the present applicant (Japanese Patent Application No. 124348/1982). 10th
This method is illustrated in FIG. In other words, the pixel pitch is 1
Then, the amount of lateral slip is finally F (So−1)≦F (So+1) as F (So
-1)>F (So+1).

Here, as shown in Fig. 8, the amount of lateral deviation is within the shift range (-
32≦S≦32), and when the contrast is sufficiently high, the amount of lateral shift can be determined with high accuracy by the above method. However, as shown in Fig. 9, even when the lateral deviation of the image is larger than the shift range, F
Since the minimum value of (S) is determined, an incorrect result of lateral shift will be produced.

This kind of case often occurs with lenses that extend a large amount, such as macro lenses and telephoto lenses. In reality, when the lateral shift becomes large, the image becomes quite blurry.
The output curve is a little more gentle than that shown in FIG.

The purpose of the present invention is to check whether the distance between two images (the amount of lateral deviation) determined by a predetermined evaluation function is a correct value corresponding to the distance from the lens to the object, and to prevent the lens from being driven by an incorrect amount of lateral deviation. An object of the present invention is to provide a focus detection device that prevents this from occurring.

[Means and effects for solving the problems] The present invention has the following features:
Two images that have passed through different optical paths are transferred to a first pixel column (column A) and a second pixel column (column B), each consisting of a plurality of photoelectric conversion elements.
In a focus detection device that calculates the distance to a subject based on the amount of shift between two images formed on them, the shift fas+1. Accurate focus detection is performed by comparing the correlation value and contrast function in S-1.

[Example] The present invention will be explained below with reference to illustrated embodiments.

Figures 1 to 4 show the results when the 20 pixels 11 to 30 of the first pixel column (hereinafter referred to as column A) are used as reference pixels and the second pixel column (hereinafter referred to as column B) is shifted. This shows the relationship between the outputs of columns A and B. The evaluation function is F(S)
=Σl"n-bn+sln 11. FIG. 1 shows the state in which p(s) becomes minimum when column B is shifted one shift at a time to the right.

Figure 2 shows shift faS. This is the image relationship when -1.

In this way S. is the shift amount before and after the image of the reference pixel row A is crossed by the image of the row congruent with it, whichever reduces the area between the images, so when the contrast function is taken as In the case of F(S)≦YC1 (1) In the case of Fig. 2, the following holds true.

Conventionally, it has been determined whether or not the object is in focus by comparing the smallest F (So) with the contrast value C or C2, but this is insufficient.

That is, as mentioned above, there are cases where the relationship in equation (1) above is satisfied even though it is not in focus, such as when the brightness of the subject image becomes complex or when the amount of lens extension is large compared to the amount of image shift. Because there is.

In such a case, if the above equation (2) is determined, the accuracy of whether or not focus is achieved will be increased.

In Figures 3 and 4, before and after column A matches column B,
When the image in row B is on the right side compared to the image in row A, P (S)
is the minimum value.

becomes. Note that as the number of pixels compared increases, C+
The contrast function may be unified to either CI or C2.

In reality, the relationship in the above equation does not strictly hold because the shape of the image becomes complex, the unevenness of output sensitivity due to the characteristics of the photoelectric conversion elements in the A and B columns, the aberration, the asymmetry of the image due to manufacturing reasons, etc. do not have. In such a case, contrast function C1,
C±α, which is obtained by adding a certain value to C2, or C2±α may be used as a new contrast function.

These are the simulation results obtained. Here, (A) is defined as (B
) cannot find the matching point of the image within the shift range, (C
) shows the results when using C2-Cl.

Below marginAs is clear from the simulation results above, if a matching point between the two images cannot be found within the shift range, F (
So)/C, or F(So)/C2 is greater than 1/2. The MIN value is smaller than 1/2, but this is because the subject is repetitive and periodic. In the phase difference method, accurate distance i! iIt&out may not be possible. There are methods to deal with this, such as using a contrast method and a phase difference method in combination, but these methods have become large. There are cases where the minimum value is smaller than 1, and the reason for this is as described above.

On the other hand, a matching point between the two images was found within the shift range (α is 1
(a sufficiently small value compared to .). C2-01
The result of (C) above is almost the same as that of (A) above.

FIG. 5 is a configuration block diagram of an electric circuit of a focus detection device showing an embodiment of the present invention.

Note that two imaging lenses L1. Many methods have already been provided for detecting the distance to an object from the correlation between two images that have passed through L2, such as those disclosed in the aforementioned Japanese Patent Laid-Open No. 59-126517 and Japanese Patent Laid-Open No. 60-247.
Since this is also explained in detail in Japanese Patent No. 210, the optical system of the present invention uses these already known means, and the explanation thereof will be omitted.

The imaging lens L1. The transmitted light of L2 is made to enter the COD line sensor 1, and the object light image that has passed through the imaging lens L1 is applied to the light receiving surface of the first photoelectric conversion element group IA forming the pixel row A, and The subject light images that have passed through the imaging lens L2 are respectively formed on the light receiving surfaces of the second photoelectric conversion element group IB forming the pixel row B. Above column A,
For example, each of the pixels on the light receiving surface of column B is formed of 64 pixels.

The drive control circuit 2 includes a COD drive circuit and an A-D converter 3.
This is a circuit for controlling the timing of COD transfer lock φ1. φ2, and an integral output signal MOS of a monitoring light-receiving element for determining the integration time of the CCD light-receiving section is input.

When this integral output signal MO3 reaches the integral level set by the drive control circuit 2, the drive control circuit 2 outputs the signal to the CCD.
An integral control signal OFG is output to the line sensor 1, and the CCD line sensor 1 thereby completes the integration. That is, the integral control signal OFG is a signal that controls the integral operation of the CCD light receiving section. Also, φ1 is a signal that transfers the charge generated in the light receiving section to the CCD transfer line, and SH is a signal that transfers the charge generated in the light receiving section to the CCD transfer line.
This is a signal for temporarily holding the CD output signal.

Further, O8 is an output signal of the CCD.

Further, RDY is a signal for notifying the CPU (central processing unit) 4 that the A-D conversion has been completed, and φ. is a reference signal for all timing signals.

In addition, the lens ROM (read-on memory) 5 stores in advance data necessary for focus detection, such as the F number of the lens and conversion coefficients for determining the amount of lens defocus from the amount of image shift. The data is sequentially input to the CPU 4. Then, based on the lens defocus amount calculated by the CPU 4, the lens drive control circuit 6 operates to move the lens to the in-focus position.

Next, the operation of the focus detection device of the present invention configured as described above will be explained with reference to the flowcharts shown in FIGS. 6A and 6B.

As shown in FIG. 6A, each pixel in the pixel columns A and B starts integration at the same time as the ranging start signal. When the integrated voltage of the monitor light receiving element reaches a predetermined value, the integrated charge of each pixel becomes C
The data are transferred to the CD transfer line, serially and sequentially A-D converted, and stored in a memory in the CPUJ allocated to each pixel.

In the following, the correlation function F(S) = Σ l”n-bn+sl
SH-22 has a total of 44 p (s) by an integer between -21 and 22.
is required.

Next, find F (S) that gives the minimum value from among these.

Now, the correlation values P(M+1) and F(M) on both sides of P(M) are
-1), when the relationship F(M+1) ≧ f'(M-1) holds true, this corresponds to the phase relationship shown in Figures 1 and 2 above, so P (M) is correctly In order for the value to correspond to the distance between and the object, the following relationship must hold. Here, 1α11°1α2 (is a sufficiently small value compared to 1.

When the above relationship does not hold, F(M) is not a value that corresponds to the distance between the lens and the object, so the display device 7 displays an out-of-focus state, and the lens is driven back and forth to Start with the integral (see Figure 6B).

Also, between the correlation values F(M+1) and r'(M-1), P
If the relationship (M+1) < P(M-1) holds, it corresponds to the phase relationship shown in Figures 3 and 4 above, so P (M) correctly corresponds to the distance between the lens and the object. In order for it to be a value, the following relationship needs to hold true, as shown in Figure 6B.

If this relationship does not hold, an out-of-focus display is displayed, the lens is moved back and forth, and the integration starts again.

On the other hand, if the correlation value is correct and corresponds to the distance from the lens to the object, interpolation calculations are then performed using the method shown in FIGS. 10 and 11 to find an accurate amount of deviation. That is, when F(M-1)≧F(M-)-i), F(M-1)
When < F (M-4-1), the deviation amount S. seek. As a result, the defocus amount is K (S-S'
). Here, K is a lens-specific coefficient, S
' is the amount of image shift when in focus, and S' is normally adjusted to zero.

FIG. 7 is a configuration block diagram of an electric circuit of a focus detection device showing another embodiment of the present invention. The CCD 11 and the A-D converter 13, whose operation timings are controlled by the drive circuit 12, are operated in accordance with the timing of the pixel output of the CCD.
The pixel output is converted from analog to digital for each pixel and stored in a predetermined memory 14.

Next, based on this pixel output data, the correlator 15 calculates F (S)-Σl an-bn,l, and the contrast function generator 16 calculates C-Σl an-a, +11, respectively. A plurality of correlation outputs p(s) generated by the correlator 15 are stored in the memory 17.

Next, the minimum value F(M) is detected from this p(s) by the minimum value detection circuit 18, and at the same time, the minimum value F(M) is detected from the same circuit 18.
) is output to the memory 17. Then, the memory 17 outputs P(M+1) and F(M-1) stored at addresses before and after F(M).

Next, in the comparator 19, I'(M+1), P(M-1
) is selected, and the contrast functions C and P(M") are divided by the divider (1) 20.

On the other hand, F (M) is divided by the contrast function C (
2) Divided by 21. Then, only when the discriminator (1) 22 and the discriminator (2) 24 determine whether or not this value is smaller than 1+α, the focus display device 26 displays that focus is possible.

In addition, in each of the above embodiments, the minimum value F (M
), P(S) does not necessarily have to be the minimum value of the correlation value.

That is, F(S) that satisfies the above normalization condition in every shift has a higher probability of being a correlation output that more accurately represents the distance to the subject than the minimum value of F(S).

However, if the above normalization is performed for all p(s), the calculation will take a long time and is not practical. For most subjects, r
Since the minimum value of '(S) is a correlation output that accurately represents the distance to the subject, and the calculation time for the minimum value of F(S) is short, we first calculate the minimum value F(M) of P(S). seek,
It is determined whether or not this satisfies the above standardization conditions. When it is determined that focusing is not possible due to the normalization of F (river), it is not the case that there is no correlation output that makes focusing possible. For example, it is also possible to perform similar normalization on the next smallest correlation output after F(M) to determine whether or not focusing is possible.

[Effects of the Invention] As described above, according to the present invention, it is possible to accurately judge whether or not the correlation value corresponding to the distance from the lens to the object is correct. It is possible to provide a focus detection device that is particularly effective for lenses with a large amount of defocus, such as telephoto lenses and macro lenses, and objects with low contrast.

[Brief explanation of drawings]

1 to 4 are output characteristic diagrams showing the relationship between the outputs of pixel rows A and B, each composed of a CCD line sensor. FIG. 5 is a focus detection device showing an embodiment of the present invention. 6A and 6B are flowcharts showing the operation of the detection device shown in FIG. 5, and FIG. 7 is an electric circuit of a focus detection device showing another embodiment of the present invention. FIGS. 8 and 9 are diagrams showing the sensor output and correlation output when shifting pixel columns A and B, respectively. FIGS. 10 and 11 are correlation diagrams showing the interpolation method. Output characteristic diagram FIG. 12 is a basic configuration diagram of an optical system of the exit pupil division method of the photographing lens.

Claims (2)

[Claims] (1) Two images that have passed through different optical paths are received by a first pixel row and a second pixel row made up of pixels of a plurality of photoelectric conversion elements, and the amount of shift between the two images formed on them is In a focus detection device that determines the distance to a subject, a correlation value F(S) representing the shift amount S with the highest correlation,
Correlation value F(S
-1) or F(S+1) becomes the desired relationship,
A focus detection device characterized by being capable of focusing. (2) Relative the fourth pixel column of the second pixel column, which has a predetermined number, to the corresponding third pixel column, which has the same number as the fourth pixel column, of the first pixel column. and a means for shifting each pixel in the third pixel column and the corresponding output of each pixel in the fourth pixel column F(S) (where S is means for determining the sum C of the absolute values of the differences in pixel outputs between adjacent pixels in a pixel column of the first pixel column in approximately the same section as the third pixel column; means for determining the minimum value F(M) of F(S), and calculating the smaller correlation value F(M′)(M ′=M
−1 or M′=M+1), [F(M′)]
Claim 1, characterized in that when /C+[F(M)]/C≦1+α (where |α| is a sufficiently small value compared to 1), focus detection is possible. focus detection device.