JPS6286317A - Focus detector - Google Patents

Abstract

PURPOSE:To simplify arithmetic processing for range finding by fixing a difference value on a prescribed value when an overflow is generated at the calculation of a difference by a difference arithmetic means. CONSTITUTION:Output from respective picture elements in a CCD image sensor 20 are successively converted into digital signals D(X) by an A/D converter 24 and inputted to a four-picture element delay circuit 28 to delay four picture elements. A subtractor 30 calculates the difference DELTAD(X) between the raw picture element data and the data D(X+4) delayed by four picture elements and outputs the difference data DELTAD(X) successively. The difference data DELTAD(X) is inputted to an overflow checking circuit 32, and if an overflow is generated in the difference data DELTAD(X), a data correcting circuit 34 fixes the DELTAD(X) on maximum data generating no overflow and outputs the fixed value. When no overflow is generated in the DELTAD(X), the data correcting circuit 34 outputs the difference data DELTAD(X) as they are.

Description

[Detailed description of the invention] (Industrial application field) The present invention relates to a focus detection device for a camera.

(Prior art) Conventionally, two images are created by forming two images of the subject light beams that have passed through two areas of a photographic lens that are symmetrical to each other with respect to the optical axis, and the distance between these two images is A camera focus detection device is known that detects the deviation of the imaging plane of the photographing lens from the expected focal plane.

The optical system of such a focus detection device has a configuration as shown in FIG. It further has a re-imaging lens 8.10 behind it, and a line sensor 12.14 based on COD, for example, is arranged on the imaging plane of each re-imaging lens 8,10.

As shown in FIG. 7, the images of each sensor 12.14-1 are located closer to each other on the optical axis 18 side in the case of so-called surface focusing, in which the image of the subject to be focused is formed in front of the intended focal plane. Conversely, in the case of rear focus, it is farther from the optical axis 18. When in focus, the distance between two corresponding points of the two images is a specific distance defined by the configuration of the optical system of the focus detection device. Therefore, in principle, the pin status can be determined by detecting the distance between the two images.

The following method is known as one of the methods for detecting this image interval. In FIG. 8, line sensors 12, 14
are made up of, for example, 10 and 16 photodiode cells a1 to alas b+ to bl[+, respectively. Further, for convenience, the numeral attached to each cell also represents the output value of each cell.

Here, in the line sensor 14, considering the sets of 10 consecutive cells, as shown in the figure, there are seven sets B + ,
B2. ..., B7 is possible. The focus state is determined by detecting which of these seven sets of images most closely matches the image of the line sensor 12.

Therefore, the following function j=1.2. −． Find for 7. J that minimizes Sj indicates the most matched set.

]O 8, -Σ l a, -b, l -=
(1) 3 i=l I l+j-1 In order to find a set that takes such a minimum value, Sj is first calculated for each set of images. Next, an operation is performed to find the minimum value from among the obtained calculation results. In this way, the focus state was detected.

For example, assume that the image of the line sensor 12 matches the image of the group B of the line sensor 14. In other words,
Cell a I-82+..., each output of alo and cell b,, b2. ..., between each output of blo,
a.

-1),, a2=bt+ “””+ ago=b+
It is assumed that the relationship o holds true. In this case, S+-1a+ btl + laz btl +
, , .

, , −+ lago boo l = O−(
2), but S is smaller than the results of similar calculations for images of sets other than set B, and is the smallest among the results of calculations for all sets of images.

However, the minimum value among the calculated values obtained above corresponds to the case where the two images most closely match, when the patterns of the two images are congruent and the sensitivity of the two line sensors is are equal to each other. If the congruence breaks down or there is a sensitivity difference between the sensors, the above minimum value will not necessarily correspond to the state of coincidence between the two images.
This becomes a focus detection error.

=4= The above will be explained using FIGS. 9(a) to 9(C).

FIG. 9(a) shows a focus detection target with step-like bright and dark areas, and the frame 20 shows two line sensors 12, 1.
Assume that the image is formed on 4. As shown in FIG. 9(b), the solid line (out+) and the broken line (out2
) respectively indicate the outputs of the line sensors 12 and 14 for the detection target shown in FIG. 9(a). The two outputs are not congruent, and a difference occurs between the outputs where they correspond to the bright area of the detection target. Note that this graph shows a case where the image on the line sensor 12 matches the image of the third group B3 portion of the line sensor 14, and the two outputs are superimposed in correspondence to the matching of the images. .

Now, if the two outputs 0LIt+ and 0LIt are congruent, the calculated value S3 shown by is O, which is the minimum compared to other calculated values. However, as shown in the graph, if the two are not congruent, the calculated value S3 will not be the minimum,
Rather, the calculated value S2 is smaller when the graph of the output 0Llt+ is shifted to the left by one cell pitch. In other words, a detection error occurs by one pitch. Now, if the ■ pitch corresponds to, for example, 30 It, then the detection error of one bit is converted to the detection error in the optical axis direction of the photographing lens, for example, about 1 mm. Such an error is of an amount that causes a practical problem in, for example, a -eye reflex camera.

On the other hand, there are other factors that disrupt the congruence between the two images. The optical system in the focus detection device is configured such that the two images formed on the upper and lower sensors 12.1.4 can be asymmetrical with respect to the optical axis 18, as shown in FIG. Furthermore, a condenser lens 6 and a reimaging lens 8. Field curvature occurs due to the aberration characteristics of the IO. Although it is possible to improve the field curvature by using an aspherical lens for the condenser lens 6 or by using a combination of multiple lenses, it is still difficult to improve the field curvature to the point where complete satisfaction is obtained. . Furthermore, two reimaging lenses 8.
It is difficult in manufacturing to configure and arrange the IO sufficiently symmetrically with respect to the optical axis of the photographing lens.

Due to the various factors mentioned above, it is impossible for the two images or the output patterns depending on each machine to be congruent. Therefore, in the conventional image comparison method as described above, the occurrence of focus detection errors cannot be avoided. do not have.

In order to improve such problems, the applicant of the present application previously proposed in Japanese Patent Laid-Open No. 60-4'914 an output signal sequence in which the output signal sequence of a photodiode cell is shifted by a predetermined number of photodiode cells. We have proposed a focus detection device that performs focus detection by calculating the difference between the two and using this difference data to obtain a calculated value similar to equation (1). In this device,
Instead of the conventional method of directly comparing the output patterns of two sensors, the phase of each output signal of each sensor is shifted by a predetermined amount in a predetermined direction.
− is generated, a difference signal corresponding to the difference between this shifted signal and the original signal before shifting is determined, and the distance between the two images is detected by comparing the thus determined difference signals of each sensor with each other. do.

For example, if the output data in FIG. 1O (b) is shifted by two pixels and a difference signal is obtained, as shown in FIG. out2'. When calculating the value of a function similar to equation (1) for this difference signal, j = 3
, which means that correct focus detection has been performed.

The reason for this improvement in focus detection accuracy is thought to be that the DC component, which is the basis of errors in the original signal, is suppressed and the AC component, which is effective for comparison, is emphasized.

(Problems to be Solved by the Invention) However, when attempting to perform arithmetic operations using such differential data using a digital arithmetic circuit, the following problems arise.

For example, when differential data is obtained using an 8-bit digital arithmetic circuit, pixel raw data (original pixel output is hereinafter referred to as this) is in the range of 0 to 255, whereas
The result of subtracting two positive numbers within the range of 8 bits is
It will range from -255 to +255. However, this cannot be expressed using 8 bits. In other words, the positive and negative numbers that can be expressed in the range of 8 bits are M
If you apply SB to the positive and negative sign bits, refer to the table below and get -128 (1000000 in binary, indicated by complement)
0B) to +127 (01111.1.11 in binary)
B). The decimal number 128 is originally expressed as 10000000 B in binary, but in a representation system that introduces negative numbers, this binary number corresponds to -128 in IO base. Similarly, decimal number -129
corresponds to 127. In this way, numbers greater than or equal to 128 and less than -129 cannot be expressed using 8 bits due to inconvenience.

In an arithmetic circuit with an 8-bit margin, the case where the result of subtraction between two positive numbers cannot be expressed completely in 8 bits will be referred to as an overflow. If data including such an overflow is used without inspection, focus detection accuracy will worsen. This is because, for example, a situation may occur where the numerical data originally 128 is treated as -128.

In order to avoid this situation, it is conceivable to increase the number of bits of the data memory from 8 to 9 and provide a sign bit, but microcomputers and memories available on the market are 8 bits or 16 bits. If you use it outside of this standard, it will be time consuming and may result in an ineffective way of using memory.
Furthermore, manufacturing a 9-bit microcomputer would be expensive.

While the pixel raw data has a positive value, the difference data has positive and negative values, so a sign bit is required. However, when performing distance measurement calculations that take these into account, the circuit configuration or calculation algorithm becomes complicated.
A large memory capacity is required, and distance measurement time becomes long. That is, it is costly, and in the case of repeated and continuous focus detection, the ability to follow the subject deteriorates.

Furthermore, as proposed by the present applicant in another application, in order to limit the difference data to 8-bit data, a method of reducing the size of data exceeding 8 bits may be considered. However, this also complicates the arithmetic processing and further increases the distance measurement calculation time, resulting in poor followability.

On the other hand, in order to prevent the occurrence of overflow in advance, it is conceivable to limit the size of the two pieces of data to be operated on. One possible way to limit the size of data is to limit the COD integration time to keep the COD output level within a certain level. For example, the integration time is limited to half of the integration time required to obtain an output level that can be expressed in 8 bits. Then, the output in that case will be at a level that can be expressed with 7 bits.

However, suppressing the signal level by such a method is undesirable because it causes deterioration of the S/N ratio for the following reasons. The image output from the COD includes a signal indicating the intensity of light incident on each cell, as well as noise that does not depend on the length of integration time. Although it is difficult to identify the cause or location of this noise, it is likely that the related circuits in the periphery include high-speed logic circuits, and that constant fluctuations in the power supply voltage, including high-frequency components, caused by the operation of these circuits may affect the signal. It is estimated that this may cause noise. In order to increase the S/N ratio against such noise, the signal component may be increased. To increase the component of the COD output signal, the COD
The goal is to ensure a sufficient integration time so that the photocurrent for optical input is integrated within the range allowed by the dynamic range of D. Therefore, from this point of view, shortening the integration time is not preferable because it lowers the S/N ratio.

Still another method is to set the input level of the AD conversion means so that the output signal is converted into a 7-bit digital value when a sufficient integration time is guaranteed.

However, in this case, the following problem occurs. As is already known, image sensors such as CCDs often output signals at a convenient level to subsequent circuits by changing the integration time depending on the brightness of the image they receive. It will be done. If the image brightness is low, a longer integration time is required, but for cameras with autofocus,
Since the image is taken hand-held, it is meaningless to increase the integration time by +3 so long that it causes an error in focus detection due to the influence of camera shake, so the integration may be terminated before the required integration time is reached. For example, an integration time that requires 400 milliseconds is cut off at 200 milliseconds.

In such a case, given an integration time of 400 milliseconds, we get
Although a signal represented by a maximum of 7 bits is obtained, since the integration is terminated in half the time, the output is represented by 6 bits. On the other hand, if the output when the necessary and sufficient integration time is given is shown as a maximum of 8-bit digital value, the output when the integration time is unavoidably limited to half as in the above case will be 7 bits. shown.

In the former case, the amount of information is 1 bit less than the latter. For highly accurate focus detection, it is preferable to have a large amount of information. For these reasons, it remains a problem to limit the AD conversion value of the output of the COD to a value smaller than the information capacity of the digital arithmetic circuit, such as 7 bits, from the beginning.

Note that in Japanese Patent Laid-Open No. 58-80607, pixel difference data for every one or more pixels is taken, and the defocus amount is calculated based on this. However, no specific explanation was given.

An object of the present invention is to perform 9-bit arithmetic operations, two's complement expression arithmetic operations, or 8-bit overflow temporary data in a focus detection device that uses the signal of the difference between the outputs of two cells located at positions different from each other by a fixed interval on a line sensor. It is an object of the present invention to provide a focus detection device that improves autofocus performance (particularly tracking ability) without performing complicated processing such as reduction.

(Means for Solving the Problems) A focus detection device according to the present invention is created from a subject light beam that has passed through a first part and a second part of a photographic lens, which sandwich the optical axis of the photographic lens. In a focus detection device that detects the relative positional relationship between the first and second images and detects the amount of deviation from the in-focus position of the photographic lens, a plurality of The number of light-receiving elements in the part that detects the second image is arranged in a row, and the number of light-receiving elements in the part that detects the second image is
The number of light-receiving elements is greater than the number of light-receiving elements in the part that detects the image (referred to as the reference part), and the number of light-receiving elements in the part that detects the second image is the same as the number of light-receiving elements that make up the reference part. A light-receiving means capable of forming a plurality of sets (referred to as comparison sections) shifted by one light-receiving element from each other, and an AD conversion means for converting the amount of light received by the light-receiving element outputted by the light-receiving means into a digital value;
The difference between the amount of light received by each light receiving element outputted by the D conversion means and the amount of light received by the light receiving element at a position shifted by a predetermined amount from that light receiving element is calculated, and when calculating the difference, the upper and lower limits of the predetermined range are calculated. If it is determined that an overflow has occurred, the difference calculation means sets the difference value to a predetermined upper limit value or lower limit value, respectively, and calculates the above-mentioned difference calculated corresponding to the light receiving element of the reference part and the light receiving element of the comparison part, respectively. A correlation calculation means for calculating a correlation function value for predetermined focus determination from the value, and a deviation amount from the in-focus position is determined from the correlation function value calculated by each of the correlation calculation means for the plurality of comparison units. and a focus determining means.

(Function) In the present invention, in the difference calculation means, when an overflow occurs during difference calculation, the difference value is fixed to a predetermined value.

(Example) Examples of the present invention will be described in the following order with reference to the accompanying drawings.

a. Structure of focus detection device 9 Focus determination C1 Flow of focus detection (a) Structure of focus detection device FIG. 1 is a block diagram showing the structure of an embodiment of the present invention. In FIG. 1, 2 is a photographing lens. Auto focus mod 2. - lens 22 is arranged behind the intended focal plane of the taking lens 2 and includes a condenser lens 6, a reimaging lens 8.10 symmetrical with respect to the optical axis of the taking lens 2, and a line sensor 12.14 in FIG. It has a COD image sensor 20 which is a single COD image sensor. The part of the CCD image sensor 0 that detects the first image is called a reference part, and the part that detects the second image is called a reference part. The reference part consists of 10 elements and the reference part consists of 16 elements. The output of each pixel of the COD image sensor 20 is sequentially converted into a digital signal D(X) by the A/D converter 24 (X = 1, 2.3, .
・). The raw pixel data D(X) is the 4-pixel delay circuit 2
8 and is delayed by 4 pixels (4 cells). The subtraction circuit 30 calculates the difference ΔD(X) between the raw pixel data D(X) and the data D(X+4) delayed by 4 pixels,
This difference data ΔD(X) is sequentially output.

ΔD(X)-D(X-1-4)-D(X)...(
4) This difference data ΔD(X) is input to the overflow check circuit 32, and it is determined whether or not an overflow has occurred. And if the difference data ΔD(
If an overflow occurs in X), data correction circuit 3
4 fixes ΔD(X) to the maximum data that does not overflow and outputs it. If no overflow occurs in ΔD(X), the data correction circuit 34 ("Difference data Δ
Output D(X) as is. The data sequentially output from the data correction circuit 34 is input to the differential data memory circuit 40 and sequentially stored.

Based on the difference data stored in the difference data memory circuit 40, the defocus calculation circuit 42 calculates the defocus amount and direction of the photographic lens 2 with respect to the planned focal plane. The output of the defocus calculation circuit 42 is input to a lens drive and display circuit 44, and is used to adjust the focus of the photographic lens 2 and to display the in-focus state.

FIG. 2 is a block diagram showing a configuration in which the circuit of the focus detection device according to the present invention shown in FIG. 1 is implemented using a microcomputer. In FIG. 2, the microcomputer 60 is a general 8-bit one-chip microcomputer (for example, Motorola Type 65).
02) with increased memory capacity is used. When a microcomputer is used, a part of the structure of the A-D converter circuit 24 and the subsequent stage of the circuit blocks shown in FIG. 1 are all formed by the microcomputer 60. Connected to the outside of the microcomputer 60 are an A-D conversion circuit 62, a switch 64 for instructing a focusing operation, a CCD drive circuit 66, a motor drive circuit 68, a motor 70, a display circuit 72, and the like. The A-D conversion circuit 62 includes a voltage comparison circuit and a D-A conversion circuit,
During A-D conversion, a time-varying digital code is given from the microcomputer 60, the corresponding analog voltage is compared with the cell output from C0D20, and when the two reach a predetermined relationship, The digital code is taken as an A-D conversion value.

The CCD drive circuit 66 generates control signals necessary for drive control of the C0D 20 from clock pulses supplied from the microcomputer 60. Motor drive circuit 6
8 is a photographic lens 2 based on the signal of the detected amount of deviation.
Power supply to the lens drive motor 70 is controlled to drive the lens to the in-focus position. The display circuit 72 displays indications such as front focus, in focus, back focus, and out of focus in the finder based on the shift amount signal. The above focus detection device has switch 6
4 is closed, the above-described shift amount detection operation is repeated and the photographic lens is moved toward the in-focus position. Note that between the CCD 20 and the A-D conversion circuit 62, for example, a circuit for removing a dark output component peculiar to a CCD or a circuit for amplifying the output signal of each cell according to the subject brightness is installed. is installed, but the explanation is omitted as it is not directly related to the present invention.The actual device is
Needless to say, it is constructed by applying various known techniques, such as changing the COD integration time depending on the intensity of light incident on the cell.

(b) Focus determination Next, the effects of this embodiment will be explained using specific numerical values.

Figures 3 (A) to (F) use Jar 1 (Figure 3 (A)), which has a black slit on a white surface as the subject, and the function I of the degree of coincidence in the distance measurement detection calculation for this.
” (D, HU).

FIG. 3(B) shows data D(X) detected in the distance measurement area 51 of the subject (FIG. 3(Δ)). Figure 3 (
C) is data obtained by taking the difference between this pixel raw data D(X) and the 4-pixel delayed data D(X+4). Reference part (
If the raw pixel data of the first image) and reference part (second image) are DΔ(X) and DB(X), respectively, ΔDA(X)=DA
(X-+-4)-DΔ(X) ...(5)ΔDB(
X)=DB(X+4.)-DB(X)...(6)
It is. Using these data, the correlation between the first image and the second image is calculated, and the -political change function r'(D) is determined, as shown in FIG. 3(E).

H'(D2ΣlΔDA(i)=ΔDB(i+j-])l
/C't = -1...(7) Here, C'7Σ"lΔDA(i)-△DA (i-+
-D l to 1 The contrast signal C° is substituted as a signal indicating the contrast of the image. In equation (7), it is used to remove the influence of contrast. If the in-focus position is selected at j where H'(D has the minimum value), the value of j will be a value related to the defocused insect, and focus detection will be possible.

8-bit arithmetic is used for data processing. It is assumed that the MSB is used for the code. However, the pixel raw data D(X) value for which the difference should be calculated is 150D(1, OOI O+,
10B) and OD (OOOOOOOOOOOB), the difference data bar +-150D (1,001
01], OB ) Also bar 150 D (011, O]
01OB). However, when calculations are performed using complement representation, which is often used in digital arithmetic processing methods, 8
For bit operations, +], 50D (1001011OB)
The current value is exactly the same as the value -+06D, and there is no difference between the values.

Also, -150D (011010], OB)Mo+10
Hard to distinguish from 6D. Therefore, even if you try to calculate the minimum value (minimum value) of H'(j) by calculating the correlation as it is to find the matching function H'(j) and find the focusing position, it will be difficult to find the minimum value (minimum value) of H'(j) due to pseudo-interpretation of the difference data. A pseudo-minimum value of
This hinders the determination of the local minimum value that should be found.

On the other hand, if we perform a 9-bit operation, +150D is (0
-106D is (11001011
.

Therefore, the method shown in FIG. 3(D), which is the main point of the present invention, is adopted. That is, to the pixel raw data D in FIG. 3(B) (X),
When taking differential data from DB(X), check whether there is an overflow or not. If there is an overflow, fix it to the largest data that does not overflow, and if there is no overflow (J, leave the data as it is) In this example, 8-bit data is considered as one piece of data, so an overflow is defined as when the difference data ΔD(X) exceeds 127 or -128.
This is the case when it becomes smaller.

That is, the difference data ΔDA(X) written in "-".

Instead of using ΔDB(X) as it is for distance measurement calculations, difference data ρX or rX defined as below is used.

ΔDA(X) > 127: 0. X-127
-128≦ΔDA(X)≦127・0. X-ΔDA(X
)ΔDA(X) <-1,28・)4 =-128・
...(8) ΔDB(X) > 127: r) (=127
−128≦ΔDB(X)≦127: rX−ΔDB(
X) ΔDB(X) < -128: rX=-128・
...(9) Using QX and rx, take the correlation between the first image and the second image, -
When the political change function H (D) is calculated, the third
The result is almost the same as in Figure (E), and the correct in-focus position can be obtained.

n Here, c-, Σ 1 di, -ρ1 + 11...(10) In this method, the data that contributes to finding the J1 correlation will be somewhat limited. However, as mentioned in 17 above, C.O.
One method is to lower the output level of D and set the circuit so that it does not overflow (unlike J, one method is to set the level so that the amount of information is rich for a large subject, and this method causes overflow. By adopting the method of this embodiment, it is possible to perform distance measurement calculations without degrading the focus detection ability and without performing complicated processing. I live without.

(c) Focus Detection Flow FIG. 4 is a flowchart showing the sequence of operations when the circuit shown in FIG. 2 is controlled using the microcomputer 60. This flowchart will be explained below.

First, in step 1, the integration operation of COD image sensor 0 is performed in nine rows. In step 2, the output is converted into a digital signal by the A/D converter 24.

Then, in step 3, this digital signal is sequentially captured as pixel raw data D(X) (X = 1, 2, 3°...)
. Next, in step 4, the difference between one data T)(X) and the pixel raw data D(X+4) shifted by 4 pixels is calculated, and this result is set as ΔD(X).

In step 5, it is determined whether this ΔD(X) is positive or negative. If ΔD(X) is positive or 0, the process proceeds to step 6, where it is determined whether there is a positive overflow. Here, we are thinking of 8-bit 1 data, so if it is thicker +3 than the maximum value of positive data +127,
In step 8, fix ρX, rX to +127, -1-1
If IJ is smaller than 27, in step 9 set rx”x (,
Substitute lΔD(X) as is. On the other hand, ΔD(X
) is negative, the process proceeds to step 7, where it is determined whether there is a negative overflow (underflow).

If it is smaller than the maximum value of 8-bit negative data -128, it is fixed at -128 in step 10, and it is larger than =128. Then, in step 11, Qx,
Store rx in memory.

Next, in step 12, it is checked whether all the data has been captured. If the data has not been captured, the process returns to step 3, but if the data has been captured, the defocus amount is calculated in step I3.

Then, in step 14, the defocus amount is output, and in step 15, the lens is driven to the focus position, and one cycle of the flow ends.

Then step 1 again for continuous autofocus.
Go back and start the next CCD Sensen integral.

FIG. 5 shows details of step 13. The microcomputer 60 finishes taking in the difference data Qx and rX used for distance measurement calculation up to step 12. Step 13
In step 21, the contrast C of the image on the basic part of the image sensor is calculated based on the calculation formula %. Subsequently, in step 22, j is shifted from 1 to 7 based on the calculation formula %, and seven correlation values H(1), H(2), . . . , H(7) are sequentially calculated. However, ρ and rk represent the digitized pixel difference output corresponding to the th pixel of the standard part and the reference part, respectively. In step 23, among these correlation values, the minimum correlation value H(jM) indicating the maximum degree of correlation is determined, and the shift position jy at that time is determined.

In the next steps 24 to 26, based on this value of jM, the minimum correlation value (maximum correlation degree) and the shift position giving it (maximum correlation shift position). First, in step 24, if it is determined that the value of jM giving the minimum correlation value is neither 1 nor 7, an interpolation calculation is performed in step 25.

This interpolation calculation calculates the shifts on both sides of jM) and position jM−]
, jM+1 are compared, and one of the following two calculation formulas is used depending on the comparison result.

When 11(JM-1)≧H(jM-+-]), (1
2) When IT(j)7-1) < 1 ((31-+-1)...(13) Here, XM is the maximum correlation shift position calculated by interpolation. In step 26, this position The correlation value, that is, the minimum correlation value (maximum correlation degree) YM by interpolation calculation is determined by the following calculation formula.

YM = H(jM) -l H(jM(-1)l((
jM-1) 1...(14) Note that when it is determined in step 24 that jM=I or 7, H(jy-1) or H(JM river) does not exist, and ``double interpolation Since this is a case where the calculation cannot be performed, proceed to step 27 and use the values of jM and r(jy) as they are as XM.
, YM.

In step 31, it is determined whether focus cannot be detected.

In the case of this embodiment, the determination is made based on whether YM/C, which is obtained by normalizing the minimum correlation value YM obtained by interpolation calculation with the contrast value C of the seven images of the reference part of the line sensor, is larger than a predetermined value A. In other words, the value of YM becomes smaller as the degree of correlation increases, and the value obtained by further removing the influence of contrast and last becomes YM/C, but if the value of YM/C is less than the predetermined value A -1- If there is, the degree of correlation is low and reliability is lacking. Therefore, when it is determined in step 31 that YM/C≧A, it is determined that focus detection is impossible, and the process returns to step 2 to perform the next optical integration. On the other hand, when it is determined that YM/C<A, the defocus amount DF is calculated in step 41. In this system, the maximum correlation shift position during focusing is
It is given by y,=4. just XM=] to XM=7
It is in the middle of That is, first, the deviation (shift) from the time of focus is determined by (XM-4). On the other hand, the amount of focus deviation (defocus@) for one software (one pitch deviation) depends on the optical system, α (e.g. 800 μm)
/Pidge), the defocus 91)F can be found from (XM-4)·α. Further, in step 42, the amount of lens driving performed by the lens driving device 40 is calculated from the defocus amount DF.

Hereinafter, returning to FIG. 4, the I-nons drive amount signal is output, the lens is moved, and the focus is automatically adjusted.

When attempting to perform continuous automatic focusing in this manner, the distance measurement calculation time is a factor that is directly related to the repetition period of automatic focusing. As in this example, it is determined whether or not the data used overflows within a predetermined level range, and if it overflows, the data is fixed at the maximum or minimum value that will not overflow, and if it does not overflow, the data is used as is. The calculation method used makes the calculation processing easier and the processing time of step I3 in the flowchart becomes faster, so quick automatic focusing with excellent subject tracking performance is possible.

Here, if it overflows, it is fixed to the maximum or minimum value before overflow, but
This may be any other value, for example 0, and a similar effect can be achieved.

(Effects of the Invention) According to the present invention, distance measurement calculation processing can be simplified, and processing algorithms can also be simplified. Furthermore, the memory capacity for holding data can be saved, and data processing functions can also be omitted.

Furthermore, the distance measurement calculation time becomes faster, and the performance of autofocus, which repeatedly performs distance measurement calculations such as in the case of continuous autofocus, is improved, especially the performance of tracking the subject.

[Brief explanation of drawings]

FIG. 1 and FIG. 2 are diagrams showing the configuration of an embodiment of the present invention, respectively. FIGS. 3A to 3F are diagrams for explaining data processing for focus determination, respectively. 4 and 5 are flowcharts of focus detection according to the present invention, respectively. FIGS. 6 to 8 are diagrams for explaining conventional focus determination methods, respectively. FIGS. 9(a) to 9(c) are diagrams for explaining the problems of the conventional focus determination method, respectively. 2. Photographing lens, 6. Condenser lens, 8.
10... Re-imaging lens, I2. I4...Line sensor, 18...Optical axis,
20 line sensor, 22... automatic focus module, 60 microcomputer. Patent applicant Minolta Camera Co., Ltd. Representative
People Patent Attorneys Previously mentioned 葆 and 2 others Figure 8 Figure 9 (c) b5b, b, b8b9blQ bll b12 No. 9
i(a) 9th City Hb>

Claims (1)

[Claims] (1) Determine the relative positional relationship between the first and second images created from the subject light beams that have passed through the first and second parts of the photographic lens, which sandwich the optical axis of the photographic lens. In a focus detection device that detects the amount of deviation of a photographic lens from its in-focus position, a portion where a large number of light receiving elements are arranged in a line to receive the first and second images, and which detects the second image. The number of light-receiving elements in the part that detects the first image is greater than the number of light-receiving elements in the part that detects the first image (referred to as the reference part), and the number of light-receiving elements in the part that detects the second image is the same as the number of light-receiving elements that make up the reference part. A light-receiving means that can form a plurality of sets of light-receiving elements arranged in a row (referred to as comparison sections) shifted by one light-receiving element from each other, and an AD conversion that converts the amount of light received by the light-receiving elements output by the light-receiving means into a digital value. calculating the difference between the amount of light received by each light receiving element outputted by the AD converting means and the amount of light received by a light receiving element at a position shifted by a predetermined amount from that light receiving element, and determining the upper limit of the predetermined range and the upper limit of the predetermined range when calculating the difference. If it is determined that an overflow has occurred with respect to the lower limit, the difference calculation means sets the difference value to a predetermined upper limit value or lower limit value, respectively, and calculates the calculated value corresponding to the light receiving element of the reference section and the light receiving element of the comparison section, respectively. A correlation calculation means for calculating a correlation function value for a predetermined focus determination from the above difference value, and a deviation from the in-focus position from the correlation function value calculated by the correlation calculation means for each of the plurality of comparison units. A focus detection device comprising a focus determination means for determining the amount of focus.