JPS6037513A - Shift amount detector - Google Patents

Abstract

PURPOSE:To detect a shift amt. with high accuracy without increasing the scale and time of calculation by forming the light image of a subject on two photoelectric element arrays and determining the shift amt. when a correlation amt. is max. by the prescribed calculation while shifting both output patterns by a prescribed amt. CONSTITUTION:The luminous flux past image forming optical system 1 passes through the light transmission region 2a of a field lens 2, is reflected by a pair of concave mirrors 4, 5 and mirrors 6, 7 in a block 3 and forms the light image of the subject on photoelectric element arrats 9, 10. The relative positions of the two light images change according to the focusing condition of the system 1. A correlation amt. C(L) is calculated from the output data (ai), (bj) (i=q-r, j-i= L) of the arrays 9, 10 in an arithmetic part 14 for the correlation. When a discriminating part 16 judges that C0 is min. from the three correlation amts. C-1, C0, C+1 with respect to the shift amts. L-1, L, L+1, the shift amt. that gives the max. correlation from the point across the axis L of abscissa is determined by an intrpolation method in an interpolation part 17. The shift amt. is thus obtd. with high accuracy.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical Field of the Invention) The present invention relates to a deviation I detection device for detecting a relative deviation of optical images projected onto a pair of photoelectric element arrays, and particularly relates to a deviation I detection device for detecting a relative deviation of an optical image projected onto a pair of photoelectric element arrays, and in particular, to The present invention relates to a deviation detection device that detects a state or a distance to a target object.

(Background of the Invention) This type of conventional deviation detection device detects the relative deviation of two optical images (1-) by connecting the photoelectric cover of one photoelectric element array to the photoelectric cover of the other photoelectric element array. The correlation between the two optical output patterns when electrically shifted by a predetermined interval with respect to the cover pattern for use is calculated as 1/, and the pattern relative shift 131 is determined from the shift amount when the correlation is maximum.

Specifically, each photoelectric element array is shown in Fig. 1(a).
Assuming that the charging elements T'h are arranged at a pitch P as shown in [,], the photoelectric cover turn obtained from the photoelectric output of each photoelectric element is shifted relatively by the pitch P as shown in Fig. The correlation between the light pattern output patterns is calculated, and the amount of correlation is calculated for each pattern shift. From the plurality of correlation amounts obtained in this way, the correlation giving the maximum correlation is selected, and a shift is performed to give this correlation l. However, since the shift leaf is in pitch units, the shift amount that gives this maximum correlation is naturally in pitch units, and as a result, the optical image deviation amount detection device described above has a detection accuracy of optical image deviation I in pitch units. The disadvantage is that it is limited to This detection device also has the following drawbacks.

As shown in FIGS. 1(h), (C), and (l), the optical image A on one photoelectric element array and the optical image B on the other photoelectric element array are relative to each other. 0 if matched
, 5P and 4.5P, respectively, are shown in FIGS. 2(a), 2(b), and 2(c), respectively, showing the correlation amount C(L) for each shift IiL. When the relative deviation between the two optical images A and H is an integral multiple of the pitch P as shown in Fig. 1(b), the shift amount corresponding to the deviation (white outline) is shown in Fig. 2(a). ), the correlation WkC
CL') is significantly smaller than the correlation amounts of other shift amounts, so the shift amount that provides the maximum correlation can be easily determined. However, the amount of deviation of the optical image is not an integral multiple of the pitch P, and in particular, as shown in FIGS. 1(c) and (d) (0,5P
If the deviation is an odd number of times, the correlation amount (indicated by a large black circle in the figure) is relatively small compared to the others but almost equal to each other, as shown in Figure 2 (b) and (c). multiple shift acids,
(indicated by an arrow). Therefore, in such a case, it is not possible to determine the shift amount that gives the maximum correlation, or even a shift amount close to it, and there is a drawback that it is impossible to detect the amount of deviation.

In addition, as shown in FIG. 2, the number of shift amounts is, for example, L=-
10, -9, -8,...8.9.10 and many (
21) would lead to a dramatic increase in circuit scale or calculation scale. Therefore, if the number of shifts is reduced, for example, L-3, -2, -1.0.1.2.3, a total of 7, then in Figure 2 (C) this shift area is the smallest value in the range. appears only at one point of shift FIL=O, so even though the relative shift amount of the optical image is 4.5P, the shift amount r,
There is also the drawback that -o is erroneously detected as the deviation at this time.

An optical image deviation detection device that solved this drawback was published in Japanese Patent Application Laid-open No. 57-
No. 45510. In principle, this device detects the amount of optical image deviation as follows.

That is, the amount of correlation is calculated while mutually shifting the photoelectric cover turns of a pair of photoelectric element arrays in units of the photoelectric element pitch P, and the function V (L) = C corresponding to the difference between these amounts of correlation is calculated.
(L-1) -C(L+1) Claim a6 le. Figure 3 (a
), city), and (C) are respectively shown in Figure 1 (t)), (C), and (d
) is plotted for each unit shift amount. When adjacent plotted points are connected by a straight line, calculate the slope of each straight line, detect the maximum slope, and find the point where the straight line with the maximum slope cuts the horizontal axis. This point becomes the shift meter that makes the maximum correlation doubtful. In this way, this detection device calculates the pitch P by calculating the function V(L) and the straight line with the maximum slope.
Since the correlation amount obtained from the Ili shift leaf is interpolated, 1. It becomes possible to accurately detect the deviation of the optical image [i2] up to m, which is smaller than the pitch P. Furthermore, as shown in FIG. 3, the straight line with the maximum slope does not appear except for those related to the maximum correlation, so if at least shift region 1 is selected to be wide fa, false detection or failure to detect will not occur.

However, like the above-mentioned detection device, this detection device also has the drawback of erroneous detection if the number of shifts is small. To illustrate, the number of shifts is, for example, L=-3, -2
, -1, 0, 1, 2, 3, a total of 7, the maximum straight line within this shift area in Figure 3(c) is the straight line passing through L=O, so the light Image shift amount is 4.5
Even though it is F, the amount of deviation is erroneously detected as zero.

Furthermore, in order to perform the above-mentioned interpolation, that is, to draw the above-mentioned straight line, four different correlation quantities are required, which has the disadvantage of increasing the circuit scale and the complexity of the calculation.

(Object of the invention) The object of the present invention is to
.

It is possible to detect the amount of deviation of the optical image up to the position, and prevent erroneous detection without unduly increasing the circuit scale or calculation scale.1]
- Possible deviation ■: Providing a quantity detection device.

A second object of the present invention is to provide a deviation amount detection device that can perform interpolation from three correlators without causing a significant increase in circuit scale or calculation scale.

The configuration of the first invention will be explained based on FIG. 4.

The optical means 100 creates a pair of optical images about the same object. The relative positions of the pair of light images change depending on the focus adjustment state of the objective lens or the distance to the target object.

The photoelectric conversion means 101 includes first and second photoelectric element arrays 102, 1.03, and this first photoelectric element array 10
The second photoelectric element array 103 receives one of the two optical images, respectively, and photoelectrically converts one of the two optical images. The correlation calculating means 104 converts the photoelectric output pattern of the first photoelectric element array 102 [3: an output pattern related to the photoelectric output obtained by appropriately processing it into the photoelectric output of the second photoelectric element array or an output obtained by appropriately processing it A predetermined amount for the pattern...
L-1, L.

Calculate the correlation amount of the Ryogawa Kabataan shifted by L + 1, - (here is an integer).

For example, Tokoname 1. The correlation amount C (
L-1), correlation amount C(L) when shifted by a predetermined amount
, the correlation amount C(L-1
) is calculated. The interpolation means 105 includes the correlation operator stage 104
Correlation amount from -C(L-*), C(L), C(L+1)
..., interpolate the extreme value correlation amount that takes the extreme value and the shift amount that gives it. The determining means 106 compares whether the L extreme value correlation amount exceeds a predetermined value. If this is possible, the interpolated shift amount that gives this extreme value is used as a signal representing the focus adjustment state or the target object distance.

The configuration of the second invention will be explained using FIG. 2.

The optical means 100, the photoelectric element arrays 102 and 103, and the correlation calculation means 104 are the same as those of the first invention. The interpolation means 107 calculates three shift amounts T7-1, L, and L+1.
The three correlation amounts C(L-1), C(L), C(
Calculate the shift amount corresponding to the extreme value between I, +1); L + -T (C(L-'1)-C(L+1))
Interpolate by. Cocoa, E = MAX (C(T-
-1') -C(L), C(L-1-1) -C(T-
)).

(Example of the invention) An embodiment of the present invention will be described below with reference to the drawings.

FIG. 6 shows an optical system of a focus detection device according to an embodiment.

In the figure, a field lens 2 is arranged near a planned imaging plane (61st-order image plane) of an imaging optical system 1 such as a photographing lens, and this field lens 2 has a rectangular light transmission area 2a in its center. However, the area other than the area 2a is a light-shielding area. The transparent block 3 in the shape of a rectangular parallelepiped is made of a high refractive index material such as glass or plastic, and the field lens 2 is attached to one end surface 3a of the block 3. A pair of concave mirrors 4.5 are provided on the other end surface 3b opposite to the one end surface 3a, and are slightly tilted in opposite directions. A pair of mirrors 6.7 are diagonally arranged at an angle of approximately 45° with a predetermined gap in between.A photoelectric conversion device 8 is arranged below the transparent block 3.This photoelectric conversion device 8 , photoelectric element arrays 9 respectively below the mirrors 6.7.
．． 10 are formed. Each photoelectric element array is composed of a plurality of linearly arranged photoelectric elements.

The light beam that has passed through the imaging optical system 1 passes through the light transmitting area 2a of the field lens 2, enters the block 3, and is reflected by the mirror 6.
7 and enters a pair of concave mirrors 4.5.

One concave mirror 4 reflects the incident light towards mirror 6, the other concave mirror 5 reflects the incident light towards mirror 7, and each reflected light passes through a mirror 6.7 to a respective photoelectric element array 9.10. reach.

In this way, a pair of object images of substantially the same object are formed in the array 9, 10J-. These subject images are
Each array 9.1 corresponds to the focal length adjustment state of the imaging optical system 1.
The relative position at 0-L changes. In the case of ζ, when in focus, the images are relatively aligned, shifted in one direction when the front bin is set, and shifted in the opposite direction when the back bin is set. Such an optical system is not limited to the above example, but may be a re-imaging optical system disclosed in JP-A-54-104859.
It may be the lenslet array optical system shown in Japanese Patent Application No. 159259, or the distance measuring optical system shown in Japanese Patent Application Laid-open No. 1212.

The photoelectric element array 9 has photoelectric outputs a1 and al of each photoelectric element.
, a, . . . a, . are generated in time series as shown in FIG.
b・, bl・b, . . . occur in chronological order as shown in Fig. 7 (can). Photoelectric element array 9 shown in FIG. 7(a)
The photoelectric output pattern 9A corresponds to the intensity distribution pattern of the optical image projected onto the array, and the same applies to the photoelectric cover turn IOA of the photoelectric element array 10 shown in FIG. 7(I). Therefore, the amount of deviation ΔX between the two light output patterns 9A and 110A corresponds to the relative amount of deviation of the optical images on each array 9.10.

In FIG. 8, - photoelectric output data a+, aS, a, . . . of each photoelectric element array 9.10 of the photoelectric conversion device 8;
and bllbl, -...bl are manually input to the processing circuit 11. This processing circuit 11 subjects the input data to non-linear transformation such as logarithm or filtering. The photoelectric output data subjected to such processing is converted into a digital value by the A/D converter 12. MC is a microcomputer and performs the functions of solid line blocks 13 to 21 shown within the dotted line block MC. The memory section 13 stores the output of the converter 2. The correlation calculation unit 14 generates a series of data a+, aS, . . . regarding the photoelectric element array 9.
a,... and a series of data b11b regarding the photoelectric element array 10! , · b, . . . are sequentially calculated by shifting the former data string by a predetermined number of data with respect to the subsequent data string. Specifically, the correlation amount C
(T,) is calculated using the following formula.

Here, L is an integer corresponding to the shift amount of the data string as described above, and the first term q and the final term r may be changed depending on the shift amount. Of course, this correlation amount C(L) is not limited to the above equation, but it is also possible to use ΣLa5-b, 7('', Σ(-a=)·h, etc.

The memory unit 15 stores three correlation amounts C (L-1) = C when the data shift amounts are increased by one data each @L-1, L, L-1-]. −,,c (r−)=c
o.

C(L+1)−C, are respectively stored in memory. The extreme value existence determining unit 16 compares the magnitude relationship between C-7 and 00, C, and cl for the stored correlation quantities c-1, co, and cl, and satisfies the conditions c-1>C6 and CI). It is determined by 61 whether CO is satisfied or not. This discriminator 1G has a discrete correlation amount C(
t, ) is plotted as shown in Figure 9(a), the correlation function F (shown by the dotted line -
It is determined whether or not t'o) has a minimum value between C-+ and C8, and when the above conditions are met, there is a possibility that a minimum value exists.

When the above-mentioned conditions are satisfied, the first interpolation section 17 calculates the above-mentioned minimum value C0 from the correlation amounts C-1, C, 1C, in the memory section 15.
1o is interpolated using the following equation.

oL=0.5 x CC-, -CI)・11) Ti:
=MAX(C, -Co%C, -C,) -+21Ce
xt = Ca l Tl, l---i3) where,
MAX (Ca, Cb) means selecting the greater of Ca and ch.

This interpolation method will be explained in detail using FIG. 9(b). In the figure, it is assumed that C,<C,<C. The maximum value C, and the minimum value C0 of these three correlation amounts are connected by a straight line t1, and a straight line m having a slope that is equal in absolute value to the slope of this straight line t1 but has an opposite sign. t-' is drawn through the intermediate value C-1. Is the intersection of these two straight lines 11.1-+ the amount of correlation? When the minimum value of tF is C., it becomes a tube. If the distance between C and C in the direction of the coordinate axis C (L) is DL, then
This distance @D, is calculated by the above equation (1).

Therefore, the minimum value Ce+u becomes C.-IT)Ll. However, since the magnitude of the correlation amount C(t) varies greatly depending on the pattern of the subject image, it is standardized so that the minimum value does not depend on the pattern of the subject image. For this purpose, using E in equation (2) of L as a normalization factor, C in equation (3) of
The value obtained by dividing 1-ITlILl by E is the normalized minimum value C...
・Suppose. In this way, if the local minimum value is made into one magnification so that it does not depend on the subject image, the correlation amount F0 that gives the minimum value of the correlation function F, that is, the maximum correlation, is almost the same as the subject image, as shown in Figure 9 (ra). It becomes an unrelated value close to zero, and the other minimum values F are considerably close to each other.

The output T′)Ll calculated by the memory unit 181f and the interpolation unit 17
E, Ce+++- is temporarily stored in memory.

The maximum correlation determining unit 19 compares the minimum value stored in the memory 18 (I7) with the reference value C1, f, and when the former is smaller than the latter, generates a comparison output.
・l is selected to be an intermediate value between the correlation amount F・ that gives the maximum function of L and the other minimum value F・.

Therefore, the determining unit 19 determines whether the minimum value F· interpolated by the interpolating unit 17 corresponds to the correlation amount F· that gives the maximum correlation.

The second interpolation unit 20 calculates the correlation Ii according to the comparison output of the maximum tn relation determination unit 19, that is, the correlation Ii where the minimum value F8 gives the maximum correlation.
'F-, the correlation +i F. (15) Formulate the corresponding shift amount L. Interpolate by -L+'-.

k is a coefficient that converts this shift amount into an amount of deviation 2 in the optical axis direction between the subject image and the predetermined imaging plane.

The memory section 21 stores the shift amount L1 that gives the maximum correlation from the second interpolation means 20, and the drive display section 22 stores the shift amount L1 that gives the maximum correlation from the second interpolation means 20, and the drive display section 22 stores the shift amount L1 that gives the maximum correlation from the second interpolation means 20. Depending on the LL11t, the front bin, wandering bin, or focus is displayed, and the imaging optical system 1 shown in FIG. 6 is driven in the direction of the focus position. Also, the second
In the interpolation unit 20, the shift amount is expressed as −
L, , is converted into the amount of deviation Z in the optical axis direction between the subject image and the predetermined image forming plane, this image plane deviation amount 2 is stored in the memory section 21, and the drive display section 22 responds to this amount Z. You can do it like this.

Note that the processing circuit 11 may perform sampling processing of input data instead of or in addition to the above-described processing.

Next, this operation will be explained using FIG. 8 and FIG. 10 showing its flowchart.

(17) (16) Outputs a1, . . . a, . of the photoelectric element array 9 and outputs hl, .
1, the data is A/D converted by an A/D converter 12, and then stored in a memory unit 13.

The microcomputer MC sets the shift scale to zero (step ■ in FIG. 10). The correlation calculation unit 14 calculates the correlation amount C(L-1) when the shift amount is L-1, L1L+1,
c(L), c (r-+-4) k is calculated, and their correlation amounts C(L-1), C(T-), C(Ll1) Each memory C-+ of the Hame memo IJ section 14, The data are stored in Co and CI, respectively (step ■). The extreme value existence determination unit 16 stores the memory C
7 and the contents of memory C0, and memories c and C
, and the conditions C, >C, TIt and C-
Determine whether or not 1λC0 is satisfied, and when it is satisfied, the memory C
It is determined that an extreme value exists between the contents of -+ and the contents of memory CI (step 2). When the above conditions are satisfied, the first interpolation unit 17 calculates the values D(,, E 1C@It , -Coat from the contents of the memories C-1, C, , C, and converts these values DL N U s Caw+- is stored in the memory unit 18 (step ■). From this interpolation) (
1B) The obtained ↑f small value Cax+ and the reference value C2°t are compared by the maximum correlation determination unit 19, and the condition C1,'< Cr
It is determined whether or not e+ is satisfied (step ■). When this condition is satisfied, the minimum value C·, I is determined to be the minimum value of the correlation function F in FIG. 9, and the second interpolation unit 20 performs the calculation r, m−(L+DL−), This value is stored in the memory section 21 (step 2). The drive display means 22 is
In accordance with this value, the imaging optical system is driven toward the in-focus position, and the focus adjustment state is displayed.

On the other hand, if the condition of step (2) is not satisfied, it is determined whether the shift scale at this time satisfies the condition L>O (step (2)). The shift scale is currently set to zero, so
The above condition L)O is not satisfied. At this time, the shift amount is −
It is updated to L+1, that is, J (step ■). This newly set shift scale is compared with a predetermined number At (step 2). This predetermined amount L+ is an integer value that determines the maximum shift amount. When the condition of this step (2) is not satisfied, the process returns to the above-mentioned step (2). After returning to step ■, when we come to step again, since L-1, the condition L>O in step ■ is satisfied, and the shifted acid is −L
1 hit-” (step O), and then step ■
Return to In this way, shift the scale from 0 to 1,
-1.2, -2.3, -3, and the maximum correlation is 4.
Adjust the amount of shift required. Shift FIL is at a predetermined value without obtaining the maximum correlation! , t, in order to express this fact, C=yl=C+el is set in step (2) and the process ends.

As the imaging optical system approaches the in-focus position, the maximum correlation decreases by 1. Gradually increasing the amount from 100 to 100 has the advantage of achieving malleable and rapid detection of the I adjustment state or distance measurement.

- In the above correlation calculation, for example, the correlation amounts c(r,) and C(L+1) when T is -0 are equal to the correlation amounts C(L-1) and C(T5) when T is -1, respectively. , when L = O, the correlation amounts C(t, -1) and C(shi) are respectively I, -1
The correlation when 1ic(L), is equal to C(I, +1).

Similarly, when L-1, C(L) and C(L+1) are respectively L
C(L-1) when =2, equal to C(L) L<, L-1
c(r-=i) when , C(L) is C- when L=-2
(T-'), which is equal to C(L+1). Despite this relationship, in the flowchart of Fig. 10, when L changes, the three correlation quantities C(T, -1), C(L), C(T, +
1) were all calculated separately. Since the computation of C(L) takes a considerable amount of time, this means an increase in computation time.

FIG. 11 shows an improved example that prevents such an increase in calculation time.

In Figure 11, at step [stock], correlation 1ic (L
-1), C(L), and C(L+1.), and store them in memories C-1, C0, and CI, respectively. (L+1)
into the memories ci, c; and the memo IJC-1, respectively.
The contents of C+1 are respectively stored in memory Co 1C+-.

After the shift mouse is incremented by 1 in step (2), in step O, the correlation amount C(L+1) is newly calculated and the contents of the memory CI are updated with this value, and in step '0, the stored memory c'' -+, c; memory C- respectively with the contents of
In addition to updating the contents of 11co, this memory C-, C
Memory C:% and C;1 are updated with the contents of I, respectively. Memory C-1, co, CI updated in this step O
The contents of are processed in step ■. Similarly, step ■
After the sign of the shift amount is inverted, the correlation amount C(L-1) is newly calculated in step 0, and this value C(
L-1), and the contents of memories Co, C';, respectively, update the contents of memories C-, , C0, C, and update the contents of memories C;, C, - with the contents of memories C-, , C. Update the contents of. Memory C,−,, updated in this step ◎
The contents of C,,C, are also processed in step (2). Other details are the same as in FIG.

In the above explanation, the shift is sequentially increased from zero to the + side and the - side alternately. Next, an example in which the shift amount is increased in one direction will be explained with reference to the flowchart of FIG. 12.

In FIG. 12, in step 2, the shift scale is set to the negative initial shift t. Correlation 1i in step [phase]
c(L) is calculated and stored in memory C-+. In step @, increase L by 1, and in step [phase], the correlation amount C
(L) and C(L+1), and store them respectively in the memory C.
o, stored in CI. At step O, memories C1 and 0
Compare each content of 0, condition C+>C.

is satisfied, the contents of memories C-8 and 00 are compared in step ①, and when the condition C-1≧C6 is satisfied, this C
Since there is a minimum value between -1 and CI, DL and E are calculated in step [phase], and the minimum value U is interpolated from the values D5, E and C0. In step 0, when this minimum value - is smaller than the reference value C...f, it is determined that this is the correlation amount that crosses the maximum correlation, and in step [phase], the shift ILm for this maximum correlation is interpolated, and the sea fence is finish.

On the other hand, when the condition CIINC1 is not satisfied in step ①, the shift amount is further increased by 1 in step [phase]. If this shift amount is smaller than the positive final value tI in step [phase], a new correlation @c(L+t) is calculated in step 2, the contents of memory C1 are updated with this, and memories C-1, C, The contents of memory C0, C
Update to the contents of I. After this, the process returns to step O. Also, when the conditions for step [phase] are not satisfied, it is clear that even if L is increased by 1, a minimum value cannot be obtained between the current shift amount [1] and L+1, so in step [phase] Increase L by 2. Also, when the conditions of step O are not satisfied, the process moves to step [phase] for the same reason. In step O, when the value after this increase is smaller than the final value 1-t, in step
After updating to the contents of I, the process returns to step O.

When L reaches the final value At in step [stock] or 0, the reference value C+al is stored in = in the memory to indicate this in step [phase], and the sea fence is ended.

In the above-mentioned embodiment, the value E used in the interpolation operation of : at the minimum value is also used as a normalization factor, which has the advantage of helping to shorten the calculation time. However, if such advantages are ignored, any value related to the contrast of the object light image can be selected as the standardization factor.
-Q B 709 The values described in the publication can also be used, and Max(C(L) -C(L-1)), Ma
x(C(L+1) -C(L-1)) or Max
(C(L)) etc. can also be used.

Next, an example in which the minimum value is not normalized will be explained based on FIG. 13.

In the shift region t, ~ifν, which of the plurality of local minimum values corresponds to the minimum value cannot be determined from the value of each local minimum value itself if the local minimum values are not normalized. I can't. Therefore, as described below, it is necessary to calculate all the minimum values in the above region and find the minimum value among them.

In FIG. 13, at step 0, an arbitrary initial value A that is sufficiently larger than the normal minimum value is stored in the memory Cml.

and an arbitrary value Aea in the memory 5°. Steps ① to ＠ are the same as steps [phase] to O in FIG. In step O, 0.5X(C-1-C
, , ) are performed and the results are stored in the memory D, and the results of the calculation Co-1+ are stored in the memory Cs**. The minimum value of the correlation amount obtained by interpolation in this way is the memory C0
*1 is stored. In step O, this memory C1
81 and each content of Cm1m are compared, and condition C*+u
(When Cmtm is satisfied, in step @, the contents of memory Cff1Ifi are changed to the contents of memory CtyI at this time, the contents of memory L are changed to the contents of memory L representing the shift amount at this time, and the contents of memory D are changed to Memo QD
The contents of L are updated respectively. Furthermore, the operation M ax (C+
Co, C-, -C,) are stored in memory E. In this way, since the memory C-(. is updated with the Cart that satisfies the conditions of step @, the smallest of the minimum values calculated so far is always stored in this memory Cml. After @ or when the conditions of step O are not satisfied, move to step [phase]. Step 0
~0 is the same as step @~0 in FIG. If the conditions of step ■ are not satisfied, move to step @. Steps O to O are the same as steps [phase] to O in FIG.

When the shift scale reaches the final value At in step (2) or 0, the minimum value Cml of the correlation amount in this shift region t, ~At is determined in step [phase]. Shift scale corresponding to,
calculates. Interpolated by 10~■.

(Effects of the Invention) As is clear from the following explanation, according to the present invention, it is possible to detect the amount of deviation with high accuracy without increasing the calculation scale or calculation time.

[Brief explanation of drawings]

FIG. 1(a) is a diagram showing the arrangement of the photoelectric element array.
Figures (b) to 1/d) are diagrams showing the relative positional relationship between two optical images, respectively, and Figures 2 (a) to 2 (c) are diagrams showing the shift amount and the shift amount by the conventional shift amount detection device. A diagram showing the relationship between correlation amounts,
3(a) to 3(c) are diagrams showing the relationship between the shift amount and the correlation amount by another conventional deviation amount detection device, FIG. 4 is a block diagram showing the configuration of the present invention, and FIG. The figure is a block diagram showing the configuration of the second invention, FIG. 6 is a perspective view showing an optical system of an embodiment of the invention, and FIGS. 7(a) and 7 et al.) are photoelectric output waveforms of the photoelectric element array. FIG. 8 is a block diagram showing the above embodiment, FIG. 9(a) is a diagram showing the relationship between the correlation amount and shift amount according to the above embodiment, and FIG. 9(b) is a diagram showing the interpolation method. 10 to 13 are flowcharts showing the operation of this embodiment. 100... Optical means 102, 103... Photoelectric element array 104 Correlation calculation means 105... Interpolation means (27) 106 - Discrimination means 107 - Interpolation means applicant 11
Hon Kogaku Kogyo Co., Ltd. Agent Takao Watanabe (28) Ogo'1 Zuko 4ε Cause 4? figure

Claims (2)

[Claims] (1) an optical means for respectively creating first and second optical images of a target object whose relative positions change according to the focus adjustment state of the objective lens or according to the distance to the target object; a photoelectric conversion means including first and second photoelectric element arrays for photoelectrically converting an optical image and a second optical image, respectively; A predetermined amount for the output pattern related to the photoelectric output of..., L-1, L, L+
1. (where L is an integer), a correlation calculation means for calculating the correlation amount of both output cover patterns shifted from time to time, and from the plurality of correlation amounts, an extremum value and the shift amount giving it. interpolation means for interpolating 1A, and determination means for determining whether or not the above extreme value exceeds a predetermined value. A deviation detection device characterized in that a shift amount is used as a signal representing a focus adjustment state or a distance as described in (1). (2) The 1'F of the target object whose relative position changes depending on the focusing state of the objective 1 lens, or depending on the distance to the target object? a photoelectric conversion means including an optical means for respectively creating a first optical image and a second optical image; and first and second photoelectric element arrays for photoelectrically converting the first optical image and the second optical image, respectively; The output pattern related to the photoelectric output of the first photoelectric element array is set by a predetermined amount to the output pattern related to the photoelectric output of the second photoelectric element...L-1, ■7, L+1,
... (Here, L is an integer.) Correlation calculation means for calculating the correlation amount of the sometimes shifted both output cover patterns, and t'11 -F specified clear L-1, Ll L+1, respectively. The correlation amounts when shifted are C(L-1'), c(1,), and C(L
+1), the shift amount that gives the extreme value correlation amount is expressed by the formula■
, ten-i-(c, (t-1)-C(L+ 1)) (where E is c(r,-H)-C(L) and C(L-1)-
It is the larger value of C(L). ) A deviation amount detection device comprising an interpolation means for interpolating according to the method.