JPH0333707A - Focus detector - Google Patents

Abstract

PURPOSE:To obtain equal accuracy in finer units with simpler constitution than a case of acquiring a correction quantity for the best correlation position by quadratic-curve approximation by finding the correction quantity by linear approximation. CONSTITUTION:There is not a large difference even when it is considered that a correlation value varies linearly while the real best correlation position is regarded as a line object nearby the best correlation position, the correction quantity (x) for the best correlation position K is found by a linear approxima tion equation from the best correlation value Hk and correlation values H(k-1) and H(k-1) before and behind it. Namely, when the correlation values are so related that H(k-1) >= H(k-1), the correction quantity (x) is found from an equa tion I, but when H(k-1) < H(k+1), the quantity is found from an equation II. Consequently, interpolative arithmetic operation is performed with the simple constitution and the same accuracy as a case wherein a quadratic curve is used is obtained.

Description

TECHNICAL FIELD The present invention relates to a focus detection device for a camera that detects a focus state by receiving a subject light beam that has passed through an objective lens, such as a photographic lens.

The first part and the second part of the conventional photographic lens sandwich the optical axis of the photographic lens.
A focus detection device has already been proposed in which the focus state is determined by detecting the correlated position of two images created by the subject light beams that have passed through the respective areas. The basic configuration of the optical system is shown in Figure 1, with a condenser lens (
4), further behind the condenser lens (4), imaging lenses (6) and (8) are arranged, and on their imaging planes, for example, CCD line sensors (10) and (12) are arranged. ing. Line sensor (lO), image on (12) (
14) and (16) are in the case of the so-called front bin, where the image of the object to be focused is formed in front of the planned focal plane.
They move toward each other toward the optical axis (18), and conversely, in the case of the rear bin, they move away from the optical axis (18). If it is in focus,
The distance between two corresponding points of the two images (4) and (16) is a specific length determined from the configuration of the optical system. Therefore, the focus state can be determined by converting the light distribution pattern of the image on the line sensor (lO), (12) into an electrical signal and determining their relative positional relationship.

The relative positional relationship is determined by comparing the two image patterns.
It can be found by calculating the correlation while shifting by pitches, but in this case, the accuracy of the amount of focus shift is rough because it is in l pitch units. Therefore, in order to achieve high precision in units smaller than the pixel pitch, it is necessary to perform a so-called interpolation calculation that predicts the true best correlation position based on the correlation values before and after the position where the best correlation was obtained. For example, it has been proposed in Japanese Patent Application Laid-open No. 75608/1983. In this device, on the premise that the above three correlation values lie on a quadratic curve, the equation is calculated and the point where the minimum extremum value is obtained is taken as the position of the true best correlation and interpolated.

Problems to be Solved However, when determining the correlation using the image signals from two line sensors as described above, the first. 1st to 2nd line senna
．． Various error factors such as variations in the characteristics of the optical system for forming the second images, errors in installation, and variations in the output characteristics of the first and second line sensors may cause the above correlation calculation results (in addition to , the above three correlation values do not necessarily lie on the ideal line, and trying to suppress the above error factors will lead to increased costs.Also, if you try to find the equation of a quadratic curve, the configuration will become complicated, etc. was occurring.

An object of the present invention is to perform the above interpolation calculation with a simple configuration,
It is an object of the present invention to provide a focus detection device that can obtain accuracy equivalent to that of a quadratic curve.

Means for Solving the Problems The present invention focuses on the fact that in the vicinity of the best correlation position, there is not much difference even if it is assumed that the correlation value changes linearly with the true best correlation position as a line symmetry, and the best correlation value H, The present invention is characterized in that the correction amount χ for the best correlation position k is determined from the correlation values 1(.-11+H(lk+ll) before and after that using a linear approximation formula.

If the correlation values before and after the action are in the relationship Hl-〇≧H(k or 1.), and if the relationship is Hfk-n < Ho+l+, then Figure 2 shows that the focus detection device according to the present invention is 2 is a diagram showing an example of the configuration of an optical system, etc. when applied to a camera. In FIG.
24), a focus plate (26), a pentaprism (28), etc. are well-known elements constituting a single-lens reflex camera. However, if the camera is configured to automatically focus using the output of the focus detection device, the shooting lens (2
2) is configured such that the focusing optical system can be driven by a lens driving device (30) including a motor. The reflecting mirror (24) has a semi-transparent central portion, and a sub-mirror (32) is provided behind it, through which a portion of the subject light is sent to the lower part of the mirror box for focus detection. The light is guided to the light receiving section (34) of the device. The light receiving section (34) includes a condenser lens (36), a reflecting mirror (38), an imaging lens group (40), a line sensor (42), and the like. The output of the line sensor (42) is processed by the signal processing circuit (44) as will be described later, and a defocus signal indicating the amount and direction of focus deviation from the in-focus position is output. Based on this defocus signal, the display device (46) displays the focus state, and the drive device (46) displays the focus state.
30) drives the photographing lens (22) to the in-focus position.

FIG. 3 is a diagram showing the optical system of the light receiving section (34), and shows the straight line (
48) indicates the optical axis of the photographing lens, and the dotted line (50) indicates a plane equivalent to the film exposure plane. Condenser lens (52
) is arranged not at the position of the exposure equivalent plane (50), but at a position far away from it by the focal length f of the condenser lens (52).

Imaging lenses (54) and (56) are arranged behind the condenser lens (52) with the optical axis (48) as the axis of symmetry, and in front of these imaging lenses is a field-limiting mask (58).
, (60) are provided. Each imaging lens (54), (5
6) On the imaging plane, there is a CCD line sensor (62),
(64) is the arrangement. Here, the condenser lens (
52) is arranged at a position away from the exposure equivalent plane (50) for the following reason. Line sensor (62), (64
), an optical system is constructed so that the object image on the exposure equivalent surface (50) is re-imaged.If a condenser lens (52) is arranged on this exposure equivalent surface (50), If there are scratches or lumps on the surface of the
This appears as an image on the line sensor and becomes noise in the original image of the object. Therefore, by removing the condenser lens (52) from the equivalent exposure plane, the above-mentioned noise can be avoided. Furthermore, when it is incorporated into a camera, it can be installed without making any major changes to the camera's optical system. Also, a mask (58),
(60) is a condenser lens (52) that receives only the subject light that passes through an aperture area corresponding to a specific aperture value, such as F5.6, out of the subject light that passes through the photographic lens.
). In this way, when various interchangeable lenses are used as photographic lenses, if the photographic lens has an open aperture value smaller than F56,
This eliminates the case where the line sensors (62), (64) receive an image in which some of the light rays are rejected by the pupil mask section of the photographic lens itself, and most commonly used interchangeable lenses can be applied.

Next, points (66), (68), and (70) on the optical axis are front focus, focus, and back focus for one object point in front of the photographing lens! fMj shows a certain image. Each statue (66), (68)
, (70) on the line sensor (62) are (72), (74), and (76), respectively, and on the line sensor (64), they are (78), (80), and (8).
2).

Figure 4 shows images (84), (86) of front focus, focus, and back focus.
), (88) are shown reimaging in the line sensor area. Re-imaging (90), (92) for the front focus image (84)
) are located in front of the light receiving surface (94) of the line sensor, and are closer to the optical axis (48) side. Focused image (8
Re-imaging (96) and (98) for 6) coincide with the light-receiving surface (94) of the line sensor, and re-imaging (100) and (102) for the rear focused image (88) coincide with the light-receiving surface (94) of the line sensor. 94) and away from the optical axis (48). Therefore, re-imaging (90
), (92) are on the light receiving surface (94) of the line sensor,
The image will be slightly blurred and stretched. In addition, the rear focus image (8
Re-imaging (100) and (102) for 8) is the light receiving surface (
94) The image above is slightly blurred and reduced in size.

Next, the relationship between the amount of movement of the image in the line sensor (62) and the amount of deviation e of the image from the in-focus position will be explained with reference to FIG. WG5. The image formed on the optical axis (48) during focusing (
Among the rays 68), consider the ray that travels parallel to the optical axis (48) after passing through the condenser lens (52). Statue (68)
The front or rear focus image (66
), (70”), then the aforementioned ray is on the exposure equivalent surface (5
At position 0), the light passes through a point (67) or (71), which is separated by g from the optical axis (48), respectively. Here, three points (68), (67), and (71) on the exposure equivalent surface (50) are used as light sources, and a condenser lens (52) and an imaging lens (5
4) By the imaging system (55) based on the above, images corresponding to the two light source numbers are formed on the line sensor (62), and the respective images are (74), (72), and (76). . Further, the magnification of the imaging system (55) is assumed to be α. If we look at FIG. 5 geometrically, the following equation holds true.

e fI a = −・・・・・・(2) If we eliminate g from these two equations, we get f.

e= h ...... (3) H In equation (3), ft/αH is a constant determined by the configuration of the imaging system, so if the amount of movement is detected, the amount of deviation e can be calculated. It will be done. However, as shown in Fig. 4, only the in-focus image is properly formed on the exposure' equivalent plane (50), and the other images are located before and after it, so it is strictly one-dimensional. The magnification α is not constant, but varies depending on the positions of the images (66) and (70) serving as light sources with respect to the imaging system (55). The magnification when focusing is α. Then, as shown in Fig. 13, in the case of the front pin, it is larger than ao, and in the case of the rear pin, α. become smaller. Furthermore, the magnification varies depending on the position of the image on the sensor surface due to aberrations such as field curvature of the optical system. Therefore, in calculating a more accurate amount of deviation, a magnification is prepared in advance according to the amount of movement and used as described later. A circuit for detecting the amount of movement and the amount of deviation e will be described below.

FIG. 6 is a diagram showing an example of the pixel configuration of the line sensors (62) and (64) in FIG. 3, where the line sensor (62) is called a reference part and the line sensor (64) is called a reference part. Pixel (
Ll) to (L!@) and (Rj) to (Rj°) are photodiodes and constitute a charge-coupled device (COD). In addition, a dummy pixel is provided in the blank space between the pixels (LlI) and (Rj), and the two line sensors (62), (
64) may be configured as one line of COD.

Furthermore, as shown in Fig. 7, the line sensor (62) and (64)
) may extend a charge transfer line (65) between them. Photodiodes (67) and (69) are for monitoring the intensity of incident light to determine the charge accumulation time of the CCD. Incidentally, this monitor photodiode may be shaped so as to fill the gap between the pixels (Ll) as shown in FIG. This makes it possible to monitor light with an intensity almost close to that of the pixel surface.

Next, in the embodiment, the image pattern at the reference part (62) of the line sensor is divided into three blocks. The first block is a pixel (Lri ~ (Ll.), the second block is a pixel (L,) ~ (Lla), the third block is a pixel (Ll.
7) to (L, ), respectively. The image pattern of each block consists of 10 pixels. Although each block has 10 pixels here, the number of pixels does not necessarily have to be the same. In focus detection, the image of each block and the image of the comparing section (64) are compared. For example, when using the image of the first block, the following comparison operation is performed.

First, the image of the portion of pixels (Rj) to (Rj°) in the reference area is symmetrically compared with the image of the first block.

The content of the comparison in this case is shown by equation (4), and the sum of the absolute values of the differences in pixel outputs in each set of pixels Ll and R8, L and R2, . . . LIO and RIO is calculated.

Hl(1) -Σ I Link(4) Next, shift by one pixel from the previous image and move the reference part (
Images of pixels (R8) to (Rj+) of 64) are compared. The processing details are shown in equation (5).

Thereafter, the comparison process shown by the following formula is performed in the same way, and the total 2 One comparison result is obtained.

Hl(3) -Σ ILh Rh+il...
... (6)H, (21) -Σ 1Lh-Rk+
i. 1(7) Now, if the image of the first professional 7 matches the image of the portion of pixel R3-Rj1, H, (Q,) will be the minimum among the 21 comparison results.

By finding the pixel area corresponding to this minimum value, the approximate focus position can be detected.

An operation similar to the comparison operation using the image of the first block is
This is done using images of the second and third blocks. The content of each comparison is generally expressed by the following formula.

H+(Q) −Σ I L h Rh−Q−+
I ... (8) H2 (12) -Σ l
L h+s-Rk+Q-+ l... (9) Q=1.2 in 22. ..., 21.

By the above comparison operation, 21 images are obtained for each block image.
In total, 63 comparison results are obtained.

Now, in the case of focusing, the image of the second block is the comparison part (62)
The optical system is configured to match the image of the pixels (Rj1) to (R20). In this way, in the case of focusing, the image of the first block matches the image of the pixels (R3) to (R+z), and the image of the third block matches the image of each part of the pixels (Rj1) to (Rza). In this case, the focus position can be detected using any block depending on the state of the image. However, in a block covered with images with low contrast, it may not be possible to identify the minimum value from the comparison results. Therefore, a plurality of blocks having a contrast of a certain value or more are selected and the focus position is detected from the comparison results corresponding to these blocks.

In addition, in the case of the front bin state, refer to Fig. 4 and refer to the reference part (
62) and the reference portion (64) coincide in the portion closer to the optical axis (48), so the image of the third block coincides with the image of the portion comprising the reference portion (64). On the other hand, in the case of the rear bin, the two images coincide in the part far from the optical axis (48), so the image of the first block is the reference part (64).
Matches the part consisting of. Therefore, in the case of out-of-focus, there is a possibility that the minimum value can be found among the comparison results for the images of the first block or wc3 block. However, if there is insufficient contrast in the image, focus detection is considered to be impossible, and the minimum value is not detected. Incidentally, in the first block and the second block, and the second block and the third block, respectively, the pixels L and Ll. and Ll, and L
l, is shared. Sharing pixels in this way,
For example, even if there is contrast in the image in the pixel area and the LID area, and contrast does not exist in other pixel areas, focus detection is possible. If pixels are not shared, if image contrast is located only at the boundary between two blocks, there will be no contrast within each block, and focus detection will become impossible.

Now, when the minimum value of the comparison result is found in any block and the matching area of the images is specified, the focus position of the image or the amount of deviation from the focus position is specified correspondingly. However, the accuracy of the amount of deviation determined through the above process is limited to the resolution of the pixel arrangement pitch. Therefore, an interpolation calculation process as described below is performed, and error factors based on the optical system of the focus detection device are corrected to improve the accuracy of the amount of deviation.

FIG. 9 (AXB) is a block circuit diagram showing a circuit configuration for processing the image pattern signal from the line sensor outlined above. This signal processing circuit has a control logic (106) that outputs control signals for the operation of the entire system including the C0D (104). CCD (104
Each pixel signal sent out in series from ) is sequentially converted into, for example, an 8-bit digital signal by a digitizing circuit (108), and each pixel signal is stored in a random access memory (110) at a predetermined address.

When the storage of the pixel signal is completed, the contrast detection circuit (112) extracts the 11th, 2nd,
Contrast of each third block Ct' C!・C1
is detected, and it is determined whether or not it is equal to or higher than a predetermined level. The contrast detection circuit z, Cs corresponds to the sum of the absolute values of the differences between the outputs of two adjacent pixels, as shown in the following equation. Note that the contrast calculation is performed without extending beyond the block area. Alternatively, the difference between the outputs of every other or every other pixel may be used.

C2-Σ I L, Ya, -Lk.

(12) C1-Σ ILh++5-Lkya, 1... (13
) The obtained contrasts C, , C, , C, are each stored in a memory (114) at a predetermined address, and the magnitude relationship is determined by a comparison circuit (116) with a predetermined level C0. If it exceeds the level C0, for example, "l" is output; if it does not, "O" is output, and the respective judgment results d, d, d for the contrast C4 Cz, Cs.

is stored in memory (120).

Next, the image comparison circuit (122) compares the image of each block with the image of the reference portion. In this case, the images of blocks whose contrast has not reached the predetermined level C0 are not compared, and only the images of the blocks whose contrast exceeds the predetermined level C0 are compared with the image of the reference portion. The contents of this comparison are as shown in equations (8), (9), and (10).

Twenty-one comparison results are obtained for each block, and these are sequentially stored in the memory (124) at predetermined child addresses. Next, the minimum values Hl (L), H2 (Qz), H3 (4s
) and their respective comparison numbers Q, , Q, , Q, are searched in the search circuit (126), and the results are stored in the memory (12
8).

Next, the standardization circuit (130) sets the minimum value Hl(
The ratio between L), Hx(cx), H3(123) and the contrast C,,C,,C3 is determined. Each is shown by the following formula.

1 t 3 These ratios mean the following. As described above, for example, when the photographic lens is at or near the in-focus position, focus detection may be possible using any of the three blocks. In such a case, the problem of block selection arises as to which block is best to adopt. Furthermore, in the case of out-of-focus, there arises the problem of determining which block should be adopted to detect the state of the front bin or the rear bin. When adopting a specific block, the minimum value H+(1゜Hz(
4z), H5(L), but this is not appropriate. In general, the contrast state of the images is not -like; for example, an image with a high contrast may be located in the region of the first block, and an image with a not-so-large contrast may be located in the other blocks. When detecting coincidence between two image patterns, it is generally advantageous to have a larger contrast. Therefore, contrast is also added to the selection of specific blocks. By the way, for example, the first
The comparison result when shifting the minimum value H, (Q□) for the block by l pixel pitch is
1), HrcQ, + l) t,: Rai f thought, t 6
゜If this minimum value Hl(L) is for the in-focus state, then Hl((21-1)AlIhaH1(12
, +1) substantially matches the contrast C1 determined by the contrast detection circuit (112). This is because the contrast C1 comparison results H'+(L l ), H+(L+
This is due to the fact that each of l) relates to the difference in output between adjacent pixels. The difference is contrast C,
is the same image, whereas the comparison results are for different images.

Since this is the case, the value N) obtained by dividing the minimum value H Expressed as a formula, however, i=1, 2, 3.

Now, if we call NHi a standardization index, we consider that the standardization index corresponding to a block that is in focus or almost in focus and has a large contrast is the smallest among the three values, and then we call this a block. stipulated in the selection criteria.

In reality, the light distribution pattern of the images of the standard part and the reference part is
Because they cannot be perfectly matched due to aberrations in the optical system and positional asymmetry of the first and second images with respect to the optical axis,
The minimum value H1 (121) never takes O. Furthermore, in the case of an out-of-focus state, the standardization index takes a relatively large value for blocks where images do not match at all. Therefore, a reference value N H is determined in advance for the standardization index, and if this value is exceeded, it is determined that focus detection is impossible.

In this way, when the minimum value of at most three standardization indexes found is smaller than the reference value NHo, the detected data Ll of the block corresponding to this minimum value is adopted as information indicating the amount of defocus. do. That is, the minimum value detection circuit (132) finds the true minimum value over a plurality of blocks. At the same time, the corresponding block is detected and the comparison number Ql that takes the minimum value H5(Q,) is stored in the memory (12
8) by the selection circuit (134). After that, the standardized minimum value NH, of the block that takes the minimum value Hh (IL) is compared with a predetermined value NHO in a subtraction circuit (136), and if NH, is smaller than NHo, proceed to the next step, and if not, proceed to the next step. It is assumed that the focus cannot be detected. Suppose now that al has been obtained for the image of the first block, and is, for example, Ql-18. This is the pixel (Ll) ~ (L
,. ) means that the images on pixels (Rj8) to (R27) match best.

In this case, the distance between the images on the two pixel areas is determined.

This interval is the interval between pixels (Ll) and (Ro). As shown in Fig. 6, if the distance between pixels (Ll) and (Rj) is 1.50 mm and the pixel pitch P is 30 μ, then D = 1.50 + 0.03X 18-2.04
(mm) (18).

Image spacing D using comparison numbers for the first block
l is expressed by the following formula.

D s=1,50+0.03a+ Similarly, for the second block, the image spacing D!
is 8 pixels shorter than the first block, so D x = 1, 50-0,03x 8 + 0.03(l
x (19) Since the third block is 8 pixels shorter than the second block, D s = 1, 50-0,03x 8 x 2 + 0.
03L...(20). Dh=1, 50-0,03(8(k-1
)+ah)...(21). The critical accuracy of the interval expressed by equation (21) corresponds to the pixel pitch P.

FIG. 10 shows an example of the comparison results for the image of block 2. The comparison number Q2 that takes the minimum value Ha (12g) is 8. If the comparison results Hz (L-1) and HzcQx+ 1 ) are not equal as shown in FIG. Comparison number az+ that takes the next smallest comparison result after L)
It exists between 1 and 9. When the position of such an intermediate point is determined, the focus detection accuracy is improved more than the pixel pitch.

Therefore, a method for determining the position of this intermediate point will be explained. Now, in Figure 1O, H1 (L-1) and HICQ□
), and draw a line whose slope is opposite to this extension line and which passes through H2(Q2+ 1 ), the point where the two intersect is considered to be the true coincidence of the two images. In this way, Hh(12h-1)≧Hhc as shown in FIG.
In the case of ah+1), the length β between the point q and the true matching point q is expressed by the following equation from the geometrical part II of the figure, and as shown in Figure 12, Hk(re-1)< Hhcak+ 1
), then .

In the circuit of Fig. 9, the compensation calculation circuit (138) and (22)
Equation (23) is calculated.

Furthermore, the equation (21) is corrected by the interpolated value β as shown in the following equation.

D' h −D h±β ・・・・・・(24
)22, the positive sign of the second term β on the right side corresponds to the case where equation (22) is used, and the negative sign corresponds to the case where equation (23) is used. As described above, the compensation calculation circuit (138)
The distance D'k between the two images in the standard part (62) and the reference part (64) is calculated from.

Next, the deviation amount e of the image of the photographing lens from the in-focus position is determined by the deviation amount calculation circuit (140) using the interval D'k. The distance between the two images when in focus. Then, the amount of image movement in FIG. 5 is calculated by the following equation.

h −−(D′ −Do) ・・・・・・ (25)
Here, h<0 indicates the front pin, and h>0 indicates the rear pin.

In the case of the imaging system shown in FIG. 5, D and -2H are used, but in reality they may differ slightly due to assembly errors, so it is preferable to set an appropriate value for Do during assembly adjustment.

Now, when the amount of movement is determined, the amount of deviation e is calculated based on equation (3).
is calculated, but the magnification σ is determined in advance according to h, for example, the first
Numerical values as shown in the table are determined experimentally and prepared in the ROM (142), and are used to calculate the deviation amount e.

Table 1 As shown above, the direction and amount of displacement of the photographic lens with respect to the subject can be determined.

FIG. 14 is a circuit diagram showing an embodiment in which a microcomputer is used in the signal processing circuit of the focus detection device of the present invention. COD (l O4) is a three-phase pulse 11+1'! from the transfer pulse generation circuit (144). + 〆, the internal transfer line is always in a data transfer state. C
In the 0D (104), the charge of each pixel is cleared by a clear pulse output from the terminal (PI3) of the microcomputer (146). Therefore, the time when the charge is cleared becomes the time when charge accumulation starts. With the start of this charge accumulation, a ramp voltage whose rate of decline varies over time in accordance with the subject brightness is output from the terminal (q,) of the CCD (104). This voltage is compared with a predetermined constant voltage Vs by a comparison circuit (148), and when it drops to this voltage, the comparison circuit (148) outputs a "high" voltage. In response to this "high" voltage, a shift pulse is output from the terminal (P+a), and in response to this, the charge accumulated in each pixel of the CCD (104) is transferred to the transfer line. CCD (104)
For this, the charge accumulation time is the period from when a clear pulse is applied to the terminal (q,) until when a shift pulse is applied to the terminal (q,). In addition to the pixels shown in FIG. 6, C0D (104) includes a plurality of pixels used as dummy pixels and a plurality of pixels for obtaining a dark output. C.O.
When a shift pulse is applied, D (104) first outputs a dummy signal and a dark signal from an output terminal (ql), and then outputs a required pixel signal. Note that when the power supply voltage Vcc changes, the output of the COD is superimposed with this change, so it is input to a circuit (150) for canceling and removing this change. This voltage fluctuation removal circuit (150) has an input (15
2) Connect the power supply voltage Vcc to resistors (154), (156)
It is given a voltage divided by , and outputs a voltage according to the difference between the two inputs. When outputting pixel signals, ccD(l O
One of the initial dark signals of the integral data output in step 4) is sampled and held by the sample and hold circuit (158), and subsequent pixel signals RoL are subtracted by the dark signal of the sample and hold circuit (158) by the subtraction circuit (160). . In this way, the pixel signal becomes one in which the voltage fluctuation component and the dark output component are removed.

The pixel signal from the subtraction circuit (160) is amplified by an amplification circuit (162) with an amplification factor depending on the brightness level. The amplification factor is controlled so that it becomes higher as the luminance level becomes lower. The brightness level is detected by the brightness level detection circuit (164) as a change in the slope voltage over a certain period of time using the slope voltage from the terminal (q8), and this change is used as a signal indicating the brightness level. . The amplified pixel signal is applied via a multiplexer (166) to an input (170) of a voltage comparison circuit (168) constituting a digitization circuit. The digitization circuit is programmed to provide the voltage comparison circuit (168), the digital-to-analog conversion circuit (172), and the 8-bit binary number to the D-A conversion circuit (172), and to store the comparison results. The microcomputer (146) is configured as, for example, a sequential comparison type A-D conversion circuit. The digitized pixel signal is stored in a memory at a predetermined address according to the pixel addresses Ri and Li. After that, the data is processed as described above,
The amount and direction of displacement of the photographic lens are detected and used to control the automatic focus adjustment of the photographic lens and display the focus state.

Now, when power supply to the microcomputer (146) is started, in response to this, the microcomputer (146)
Step 6) moves on to the COD initialization program. Before focus detection starts, COD (104)
Charges are accumulated in the transfer lines and pixels above the normal pixel signal level, but before extracting the pixel signal,
This unnecessary charge is cleared from the transfer line and pixels.

This clearing operation is the initialization of COD.

In this initialization, a clock pulse with a cycle shorter than that during normal pixel signal transfer (for example, normal l/I 6) is applied to the COD, and the transfer operation faster than normal is repeated multiple times (for example, 10 times). This leaves the transfer line empty. In parallel with this, pixel clearing is also performed. In this case, no pixel signal capture operation is performed. The transfer pulse generation circuit (144) generates transfer pulses 11,1'2.1s using a clock pulse of a constant period from the terminal (P, ) of the microcomputer (146).

The transfer pulse, which has a shorter cycle than normal, is generated when the flip 70 knob (176) is in the reset state and its output is 1 high.
voltage, the frequency division ratio of the clock pulse is changed by a predetermined value within the transfer pulse generation circuit (144) according to this "high" voltage.Flip flow rough' (176) is reset by a pixel charge clear pulse from the microcomputer (146) and set by a shift pulse.The shift pulse also puts the transfer pulse generation circuit (144) into a state where it generates a normal transfer pulse. , CCD
(104) is defined as the time from when the charge clear pulse is generated to when the shift pulse is generated as the charge accumulation time, and during this period, the transfer pulse generation circuit (144) outputs a transfer pulse with a shorter cycle than normal. However, since the signal outputted from the CCD (104) via the transfer line during the charge accumulation period is treated as an unnecessary signal, no problem occurs even if the transfer pulse becomes faster.

Now, when a predetermined number of transfer cycles are completed as an initialization operation, the microcomputer (146) moves to the program for focus detection described above. First, when a clear pulse is output, COD (l O4) starts accumulating charge. At the same time, the terminal (q2) of C0D (104) outputs a ramp voltage that drops from a predetermined voltage at a rate according to the subject brightness, and when this voltage drops to a predetermined level Vs, the voltage comparison circuit (148) The output level of is reversed from “low” to “high” voltage. This high voltage is used as an interrupt signal, and when the microcomputer (146) receives an interrupt, it outputs a shift pulse from the terminal (P,).The shift pulse causes CCI) (1
The charges accumulated in each pixel of 04) are transferred in parallel to the output terminal (ql), and then transferred in series to the output terminal
are sequentially output as voltage signals. This voltage signal is digitized as described above and stored in a predetermined memory. When the pixel signal capture is completed, the terminal (P,
), for example, a "high" voltage signal is temporarily output, and in response, the multiplexer (166) activates the constant voltage circuit (
178) is selected and outputted, this constant voltage is digitized by a digitizing circuit (108), and taken into a predetermined memory. As mentioned above, the distance between the two images formed on the standard part and the reference part during focusing does not match the design value due to assembly errors in the optical system, so this data is used to correct this error. used as. The constant voltage circuit (178) is composed of a constant current circuit (180) and a semi-fixed resistor (182), and the semi-fixed resistor (182) is adjusted in the adjustment process of the focus detection device to set accurate image distance data. It will be done.

FIG. 15 is a 70-chart showing the flow of the operation of the focus detection device described above.

Effects As described above, according to the present invention, since the correction amount for the best correlation position is determined by linear approximation, the same accuracy can be obtained in finer units compared to the case where the correction amount is determined by quadratic curve approximation. Moreover, it can be obtained with a simple configuration.

[Brief explanation of drawings]

Fig. 1 is a diagram showing a conventional example of an optical system of a focus detection device, Fig. 2 is a diagram showing an example of the arrangement of the focus detection device of the present invention in a camera, and Fig. M3 is a diagram showing an optical system of the focus detection device of the present invention. 4 is a diagram showing the image formation state by the optical system of the focus detection device of the present invention, and FIG. 5 is a diagram showing the amount of focus shift in the optical system of the focus detection device of the present invention and the line sensor. Figures 6, 7, and 8 are diagrams showing the relationship between the upper image and the amount of movement, and Figures 6, 7, and 8 are diagrams showing examples of the pixel configuration of the line sensor of the focus detection device according to the present invention, and Figure 9 (AXB) is , a block circuit diagram showing the configuration of the signal processing circuit of the focus detection device according to the present invention, FIG. 10, 1.1ffi and 1
FIG. 2 is a graph for explaining the operation of the signal processing circuit, FIG. 13 is a graph showing the magnification of the optical system of the focus detection device according to the present invention, and FIG. 14 is the signal processing circuit for the focus detection device according to the present invention. FIG. 15 is a block circuit diagram in the case where a microcomputer is used. FIG. 15 is a 70-chart showing the flow of operation of the signal processing circuit. 2.22...Photographing lens, 12.14.62,64.
104...Line sensor (CCD), 4.36°52
...Condenser lens, 6,40.54.56...
Imaging lens, 67.69...Subject brightness monitor photodiode

Claims (1)

[Scope of Claims] A focus detection device that detects the focus state of an objective lens by detecting the correlation between first and second images formed by subject light beams that have passed through different parts of the objective lens. A first image signal L_i (where i=1, 2, . . . ,
m), and a second image signal R_j() which receives a second image and outputs a second image signal R_j(
However, the second output which outputs j=1, 2,..., n: m<n)
The correlation value Hl between the line sensor and the first image signal and the second image signal in units of pixel pitch of the line sensor is expressed by the formula Hl=
■｜L_i-R_(_i_+_l_-_1)｜(However, l
= 1, 2, ..., nm+1), the correlation value H_k at position k where the best correlation was obtained among the correlation values Hl, and the correlation values H_(_k_+
_1_), H_(_k_-_1_) to calculate the correction value χ for the position k at which the best correlation is obtained from the following linear approximation formula in units less than the pixel pitch; A focus detection device characterized by comprising: determining means for determining the amount of defocus based on the position k and the correction value χ: χ=1/2(H_(_k_−_1_)−H_(_k_+
_1_))/(H_(_k_-_1_)-H_k) However, H_(_k_-_1_) ≥ H_(_k_+_1_) or χ=1/2(H_(_k_+_1_)-H_(_k_-
＿１＿))／(H＿(＿k＿＋＿１＿)−H＿k)However, H＿(＿k＿−＿１＿)＜H＿(＿k＿＋＿１＿)
It is.