JPS6236206B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a focus detection device that detects the in-focus state of an objective lens (imaging optical system).

Conventionally, this type of device guides a pair of optical images formed by two light beams that have passed through spatially different parts of an objective lens onto a pair of photoelectric element arrays, compares and processes the outputs of the pair of photoelectric element arrays, and then synthesizes the images. Various methods have been proposed to detect the focused state.

For example, Japanese Patent Application Laid-Open No. 142512/1983 describes a method of detecting the in-focus state using the sum of the differential outputs of the pair of photoelectric element array outputs and the contrast. In this conventional example, a focus detection device is described by taking advantage of the fact that the relative deviation of the pair of optical images changes depending on the in-focus state, and that the contrast is maximum when the images are in focus. According to this method, the detection output curve h ₃ rises sharply at the in-focus position (as shown in FIG. 5 of JP-A-52-142512), and the in-focus range is determined by setting the threshold value ε ₀ . ing.

However, in this conventional example, since this detection output curve is affected by contrast,
Although it functions properly for a specific subject, for example, when the subject has a different amount of contrast, the level of this detection output curve _h3 changes (that is, shifts in the vertical axis direction in Fig. 5). threshold ε
If ₀ is kept constant, there is a drawback that good results cannot be obtained. Furthermore, although the arithmetic processing of this conventional example can determine the in-focus position and the vicinity of the in-focus state, it has a drawback that it cannot determine the direction or amount of image shift, such as front focus or back focus.

SUMMARY OF THE INVENTION An object of the present invention is to solve these drawbacks and provide a stable focus detection device that is not affected by the contrast of a subject.

In order to achieve the above object, the present invention is provided with a contrast calculation means for calculating a contrast amount signal indicating the sharpness of an image, and a correlation calculation means for outputting a correlation amount signal that changes depending on the relative displacement of two images. , further comprising first determining means for comparing the contrast amount signal with a first predetermined level, and second determining means for comparing the correlation amount signal with a second predetermined level in a form in which the influence of the contrast amount signal is removed. The focus detection device is configured to ensure that the sharpness of the image of the object to be focused is the amount necessary for focus detection by comparing the contrast amount signal with the first predetermined level; The detection signal of the focus detection calculation that contains many errors due to the extremely low contrast of the image (that is, the uncertain detection signal) is
An effect that can be known in advance and can be dealt with is obtained, and further, by comparing the correlation amount signal with a second predetermined level in a form in which the influence of the contrast amount signal is removed, it is possible to eliminate the influence of the contrast of the subject image. The effect is that the range in which the signal obtained by the focus detection calculation is valid can be appropriately identified.

Furthermore, in the embodiment of the present invention, the amount and direction of deviation between the two images are detected by focus detection calculation, and the detection can be performed with high precision at least in the vicinity of the coincidence of the two images.

In FIG. 1, a field lens 2 is provided on a fixed focal plane of an objective lens 1 or a plane conjugate thereto. In a camera, a film is placed on a fixed focal plane, so the field lens 2 shunts the optical path of the objective lens 1 and is provided in the shunted optical path. Regarding the re-imaging lenses 3 and 4, photoelectric element arrays 5 and 6 are provided at positions conjugate with the fixed focal plane of the field lens 2, that is, the lens 1, respectively. In this example, each array 5, 6 consists of eight photoelectric elements _P1 to _P8 , _P1 ' to _P8 '. On the object that the objective lens should focus on,
When focused, the optical image of the object formed by the objective lens 1 and the reimaging lenses 3, 4 on the photoelectric element arrays 5, 6, respectively, and the corresponding array 5,
The positional relationship between the re-imaging lenses 3 and 4 and the arrays 5 and 6 is determined so that the positional relationship between the re-imaging lenses 3 and 4 and the arrays 5 and 6 is the same. Therefore, in the focused state, a pair of positionally corresponding photoelectric elements P ₁ and P ₁ ' of the array 56...
The intensities of light incident on P ₈ and P ₈ ′ are equal. Also, the image of the object formed by the objective lens 1 is transmitted to the field lens 2.
When formed in front of (front pin) array 5
The upper image moves downwards, while the image on array 6 moves upwards. Conversely, when the image by the objective lens 1 is formed behind the field lens 2 (rear focus), the images on the arrays 5 and 6 move in the opposite direction to the front focus. Each photoelectric element P ₁ of array 5
The photoelectric output of ~ _P8 is linearly amplified or, respectively,
The output terminal 5 of the array 5 is objectively amplified and the electrical output υ ₁ to _{υ 8} related to the photoelectric output is
Output from a to 5h. Photoelectric element of array 6
The photoelectric outputs of P ₁ ′ to P ₈ ′ are similar, and the related electrical outputs υ ₁ ′ to υ ₈ ′ are outputted from the output terminals 6a to 6h.

Next, this related electrical output υ ₁ ~ υ ₈ , υ ₁ ′ ~ υ
₈ ′ processing will be explained. In FIG. 2, the spatial frequency component extraction circuit 7 generates a first electrical signal V ₁ representing a specific first spatial frequency component of the optical image on the array 5;
A second electric signal V ₂ representing a spatial frequency component with a spatial period of 1/2 of the above-mentioned related electric output υ ₁ to υ
Extract from ₈ . The second spatial frequency component may have a spatial period different from the first spatial frequency component, and is not limited to the above example. When the optical image on the array 5 is displaced in the arrangement direction of the elements, this first electric signal V ₁ contains phase information φ ₁ that changes in a fixed relationship according to the displacement, and the magnitude of the extracted spatial frequency component. size information _γ1 representing the size. The second electric signal is similar and includes phase information φ ₂ and magnitude information γ ₂ . The other spatial frequency component extraction circuit 8
is the same as the circuit 7, and extracts the first and second spatial frequency components of the optical image on the array 6 from the associated electrical outputs υ ₁ ′ to υ ₈ ′ and represents them respectively. Create a second electrical signal. This first,
The second electrical signals include phase information φ ₁ ′, φ ₂ ′ and magnitude information γ ₁ ′, γ ₂ ′, respectively.

The principle of this extraction circuit will be explained with reference to FIG.
In the same figure, the associated electrical outputs υ of photoelectric elements P ₁ to _{P 8}
The vectorization circuit 9 sequentially converts ₁ to υ ₈ into vectors exp (i2π×0/8
), exp(i2π×1/8)...multiplyed by exp(i2π×7/8) and converted to a vector. Addition circuit 10 adds the outputs of vectorization circuit 9. Therefore, the adder circuit 10
The output V is as follows.

This output υ ₁ includes information of a spatial frequency component whose spatial period is the spatial length d in the arrangement direction of the photoelectric elements P ₁ to _{P 8} of the array. As demonstrated in Japanese Patent Application No. 53-12144, the phase φ of this output V changes when the optical image on the array is displaced. Further, the magnitude γ of the output V represents the magnitude of the extracted spatial frequency component in the optical image.

If the above-mentioned output V is the first electrical signal, the second electrical signal regarding the spatial frequency component with a different spatial period is
To obtain the electrical signal, proceed as follows. In the vectorization circuit 9, the related electrical outputs υ ₁ to υ
By multiplying ₄ , υ ₅ to υ ₈ by vectors exP (i2π x 0/4) to exP (i2π x 3/4), respectively, the spatial frequency with a period equal to the length d/2 of the four photoelectric elements is obtained. ingredients are obtained.

Specific circuit examples of the arrays 5 and 6 and the spatial frequency extraction circuits 7 and 8 are shown in FIGS. 4 and 5.

In FIG. 4, the photocurrents of the photodiodes P ₁ to P ₈ and P ₁ ' to _{P 8} ' of the arrays 5 and 6 flow through the operational amplifier and its feedback transistor. A related electric output υ ₁ proportional to the logarithm of the incident light intensity is generated by the logarithmic conversion circuits 12a to 12h, 13a to 13h consisting of diodes.
〜υ ₈ , υ ₁ ′〜υ ₈ ′. Note that the feedback transistors and diodes of the logarithmic conversion circuits 12b to 12h and 13a to 13h are not shown in the figure.
The circuit 15 has associated electrical outputs υ ₁ to υ ₈ , υ ₁ ′ to υ
Feedback is applied so that the average value of ₈ ' becomes zero.
The purpose of setting this average value to zero is to reduce the influence of errors in the value of the vector multiplied by the vectorization circuit, specifically, errors in the input resistances of the left-hand amplifiers 16 to 23.

Next, in FIG. 5 showing the extraction circuit, terminal 5a
~5h, 6a~6h are vectoring circuits connected to terminals 5a _~ 5h, 6a _{~ 6h} with the same symbols in FIG.
₈ ′ is the vector exP(i2π×0/8) ~ exP(i2π×7/
8) in the form of its x and y components. The differential amplifier 16 multiplies and adds the n components of the above-mentioned vector to the related electrical outputs υ ₁ to υ ₈ .
The value of each input resistance is proportional to the reciprocal of the x component to be multiplied. Thus the differential amplifier 1
The output of 6 becomes the x component of the first electrical signal V ₁ regarding the spatial frequency component of the spatial period d.

The differential amplifier 17 multiplies the related electrical outputs υ ₁ to _{υ 8} by the y component of the vector described above and adds the results. This output corresponds to the y component of the first electrical signal _V1 . Also, the differential amplifiers 18 and 19 produce the x and y components of the first electrical output V ₁ ' in exactly the same way as the amplifiers 17 and 18 for the associated electrical outputs υ ₁ ' to υ ₈ ' of the array 6. Also, the differential amplifier 2
0,21 is x of the second electrical signal V ₂ with respect to the spatial frequency component of a spatial period of d/2 for the array 5
component and y component, respectively, and outputs the differential amplifier 2.
2 and 23 are second electrical signals for array 6;
The x and y components of V ₂ ' are output respectively.

Multipliers 24, 26, 28, 30 multiply the outputs of differential amplifiers 16, 18, 20, 22 by the AC output coswt from circuit 32, and multipliers 25, 27, 2
9, 31 are differential amplifiers 17, 19, 21, 23
The output of is multiplied by the AC output sinwt from the circuit 32. Adders 33, 34, 35, 36 are multipliers 24, 25, 26, 27, 28, 29, 30, respectively.
and the output of 31 are added. These adders 33,
The AC outputs of 34, 35, and 36 correspond to the first electrical signals V ₁ , V ₁ ′ and the second electrical signals V ₂ , V ₂ ′, respectively, and their phases are the above-mentioned phases φ ₁ , φ 1 _′ , φ ₂ , φ ₂ ', and their outputs are connected to rectifier circuits 37, 38, 39,
40 becomes the size information γ ₁ , γ ₁ ′, γ ₂ , and γ ₂ ′, respectively.

Again in FIG. 2, the phase difference φ ₁ −φ ₁ ′ between the first electric signals V ₁ and V ₁ ′ of the spatial frequency component extraction circuits 7 and 8 obtained in this way is determined by the fact that the objective lens is aligned as shown in FIG. 6 a. It is zero at the focused position, positive at the front focused position, and negative at the rear focused position, and the magnitude of the difference increases depending on the amount of deviation from the focused position.

The same holds true for the phase difference φ ₂ -φ ₂ ' between the second electrical signals V ₂ and V ₂ ', as shown in FIG. 6b.

For example, the second electrical signals V ₂ and V ₂ ' form one cycle of four pixels as explained above, and therefore the width of one pixel corresponds to a phase difference of 90°. 7th
The phase difference detection circuit shown in the figure can easily divide this 90° phase difference into several tenths, so Φ ₂ - Φ
If a signal of ₂ ' is used, it is possible to detect a relative displacement of more than a few tenths of a pixel width, and the detection accuracy is high.

That is, as shown in FIG. 6, focus detection calculation is performed that can accurately detect the displacement of the two images, including the direction, although the effective range of application is limited to the vicinity of the state where the two images match.

Next, although the detection accuracy is rough, a method for identifying the vicinity of the state where the two images match from other states will be described in detail below. In order to achieve this purpose, the present invention provides two
Correlation calculation means for outputting a correlation amount signal regarding the relative displacement of the image, and contrast calculation means for outputting a contrast amount signal indicating the sharpness of the image,
By creating a standard correlation output by normalizing the correlation amount signal with the contrast amount signal and comparing this with a predetermined threshold, a method is used to effectively identify the vicinity of the matching position of two images for any subject. . The correlation detection unit 41 in FIG. 2 performs this process. However, as is clear from FIGS. 6a and 6b, the phase differences φ ₁ - φ ₁ ' and φ ₂ - φ ₂ ' become zero even when there is a large shift from the in-focus position to the front focus or rear focus position. Therefore, if focus detection is performed only from this phase difference, there is a risk of errors. In order to prevent this, a correlation detection section 41 is provided that detects that the objective lens is near the in-focus position. This correlation detection section 41 uses a standard correlation function Calculate. The numerator of this standard correlation function 1 becomes small when the brightness distribution of the object is almost uniform, and the denominator also becomes small accordingly, so this standard correlation function is independent of the brightness distribution and depends on the focal position of the objective lens. It has become standardized. The numerator of this standard correlation function represents the cross-correlation function for a pair of optical images, and the denominator is the autocorrelation function of the optical images, that is, the contrast (sharpness) of the optical images.
represents. To explain in detail, at the in-focus position, υi=
Since υ′i, the numerator becomes zero and = 0, and the object is in the rear focus or front focus such that the image on array 5 is shifted by one photoelectric element relative to the image on array 6. υ ₂ ′= regardless of the luminance distribution of
υ ₁ , υ ₃ ′=υ ₂ ,…υ ₈ ′=υ ₇ or υ ₂ = υ
₁ ′, υ ₃ = υ ₂ ′…υ ₈ = υ ₇ ′ hold, so it becomes =1.

In this way, this standard correlation function is standardized at three points: the position α where the correlation is maximum, and the positions β and γ where the relative displacement of the image corresponds to ±1 pixel, as shown in FIG. 6C. That is, the range of around ±1 pixel corresponds to the vicinity of the state where the pair of optical images match. Of course, such a standardized correlation function is not limited to this, and various other types can be considered. For example, the denominator of the standard correlation function can be any quantity as long as it corresponds to the contrast of the optical image, and by substituting this denominator with the sum of the magnitude information r ₁ , r ₂ , r ₃ , r ₄ of the spatial frequency components, But that's fine.

That is, the standard correlation function is obtained by normalizing the amount of correlation between two images by the amount corresponding to the contrast.

This completes the explanation of the method of identifying the vicinity of the state where two images match, although the accuracy is rough, from other states, and the seventh
The control unit 50 in FIG. 2 will be explained based on the circuit of the embodiment shown in the figure.

The comparator 51 receives 1/2 of the sum of magnitude information γ ₁ , γ ₁ ' regarding the first spatial frequency component with the spatial period d, and magnitude information γ ₂ , γ regarding the second spatial frequency component with the spatial period d/2. ₂ ', and when the former is large, that is, when the optical image contains many first spatial frequency components, an "H" level output is generated, and the FET52 regarding the phases φ ₁ and φ ₁ ' becomes conductive. On the other hand, when the latter is large, an "L" level output is generated and the phase φ
_2. Make FET54 related to φ ₂ ' conductive. Note that the reason why the sum of γ ₁ and γ ₁ ' is multiplied by 1/2 here is that the phase information regarding the second spatial frequency component is more accurate than that of the first spatial frequency component. . When FET52 is on, AC signals φ ₁ and φ
₁ ', and when the other FET54 is on, the AC signal φ ₂ ,
Each of the waves φ ₂ ' is converted into a rectangular wave by a waveform shaping circuit 55 and then inputted to a D-flip-flop 56 and an exclusive OR circuit 57. The D-flip-flop 56 adjusts the phase difference between the two AC signals φ ₁ -φ ₁ ' or φ ₂
- Determine whether φ ₂ ′ is positive or negative. As mentioned above, the positive and negative numbers represent the front pin or the rear pin. The exclusive OR circuit 57 measures the absolute value or magnitude of the phase difference. This size represents the amount of deviation from the in-focus position. Therefore, from the output of the flip-flop 56 and the output of the circuit 57, the in-focus state, front focus, rear focus, and the degree thereof can be determined.

Comparator 58 compares the standard correlation function output with reference level _V1 , and generates an "L" level output when the former is low. This reference level V ₁ is as shown in FIG. 6c, and in the range where the correlation function output is below this level, the phase difference φ ₁ -φ ₁ ', φ ₂ -φ ₂ as shown in FIGS. 6a and b. ′ is always in the normal range l ₁ , l ₂
It has been selected to ensure that In this way, when the output of the comparator 58 is at the "L" level, the objective lens is near focus, and the output of the D-flip-flop 56 and the output of the exclusive OR circuit 57 are normal focus detection signals. Guaranteed.

There is an object such as a blackboard whose luminance distribution is almost uniform and the optical image on the arrays 5 and 6 contains almost no first spatial frequency component or second spatial frequency component. For such an object, the phase differences φ ₁ −φ ₁ ′, φ
₂ -φ ₂ ' is greatly affected by noise and the like, and no longer faithfully represents changes in the relative positions of the optical images on the arrays 5 and 6. Therefore, this phase difference is no longer an accurate focus detection signal for the objective lens. Therefore, when such first and second spatial frequency components are hardly included, the size information γ ₁ , γ ₁ ′, γ ₂ , γ ₂ ′ is naturally becoming smaller, so the comparator circuits 60 and 61 Information γ ₁ , γ
₁ ' is above a certain level V ₂ , and similarly the comparison circuits 62 and 63 detect whether the information γ ₂ and γ ₂ ' are above a certain level V ₃ , and both are above a certain level. The output of the OR circuit 64 is set to "L" level only in certain cases. That is, the "L" level of the output of this OR circuit 64 is the flip-flop 5.
6 and the output of circuit 57 are correct.

Furthermore, if the denominator of the correlation output is very small, the correlation output will likely contain many errors, so in this case it is inappropriate to use the output of the comparator 58 as a signal representing the vicinity of focus. Therefore, the comparator 65 detects whether the denominator is greater than a certain level _V4 . And the comparator 65 is
When the denominator is smaller than _V4 , it becomes "H" level, and this "H" level means that the output of comparator 58 indicating whether or not the focus is near may be incorrect. When the output of the OR circuit 66 is at "H" level, the phase difference φ ₁ -φ ₁ ', φ ₂ -
This indicates that at least one of φ ₂ ' or the standard correlation output is incorrect.

In this example, two phase differences φ ₁ −φ ₁ ′ and φ ₂ −φ ₂ ′ were created to improve detection accuracy.
Of course, there may be only one.

Furthermore, in the example shown in Fig. 4, the outputs from the photodiodes are taken out in parallel, but for example,
When extracting data in time series, if AGC is applied so that the charge accumulation time of the photodiode becomes constant γ ₁ + γ ₁ ′ + γ ₂ + γ ₂ ′=constant, the correlation output Σ|
υ _o −υ′ _o | becomes standardized.

As described above, according to the present invention, by comparing the contrast amount signal of the contrast calculation means with the first predetermined level, it is guaranteed that the object image to be focused has the sharpness necessary for focus detection. By comparing the correlation amount signal with the second predetermined level in a manner that does not depend on the contrast of the optical image, the correlation amount signal can be properly adjusted to the second level without being affected by the brightness distribution of the subject image.
Identify the neighborhood of the state where the image matches from other states,
The amount of deviation between the two images, including the direction, is precisely calculated in the vicinity of the two images matching, making it possible to determine front and rear focus, with high detection accuracy, and a stable focus detection device that does not malfunction regardless of the subject. can be obtained.

[Brief explanation of the drawing]

FIG. 1 is a layout diagram of an optical system according to an embodiment of the present invention.
FIG. 2 is a block diagram of a signal processing system according to an embodiment, FIG. 3 is a block diagram showing the principle of a spatial frequency component extraction circuit, FIG. 4 is a circuit diagram showing a specific example of a photoelectric element array, and FIG. The figure is a circuit diagram showing a specific circuit example of a spatial frequency component extraction circuit, FIGS. 6a and 6b are graphs of phase difference output, and FIG. 6c is a graph of correlation output. FIG. 7 is a circuit diagram showing a specific circuit example of the control section. Explanation of symbols of main parts, 1...Objective lens,
5, 6...Photoelectric element array, 7, 8...Spatial frequency component extraction circuit, 41...Correlation detection unit.

Claims (1)

[Scope of Claims] 1. A photoelectric conversion means for photoelectrically converting a pair of optical images created from the light beams by allowing light from the same part of the subject to pass through an optical aperture having parallax through different paths, respectively; a focus detection means for performing focus detection calculation based on the relative displacement of the pair of light images on the conversion means; a contrast amount calculation means for outputting a contrast amount signal indicating the sharpness of the light image; correlation calculation means for outputting a correlation amount signal related to target displacement; first determination means for determining that the contrast amount signal of the contrast calculation means exceeds a first predetermined level; and a second determining means for determining that the signal exceeds a second predetermined level with the influence of the signal removed, and the determination results of the first determining means and the second determining means are respectively the first and second predetermined levels. and a second predetermined level, the focus detection calculation is determined to be appropriate. 2. According to claim 1, the focus detection calculation of the focus detection means calculates the magnitude and direction of the relative displacement at least in the vicinity of a state where the pair of optical images match. The focus detection device described.