JPH0320707A - Automatic focusing device and detection of focusing position - Google Patents

Abstract

PURPOSE:To improve the S/N of a focus signal and suppress the maximal point, and to perform fast, high-accuracy focus adjustment by adding the output amplitude value of the frequency component of an image signal and generating a focus signal, and filtering this signal. CONSTITUTION:An image formed by a photographic optical system 31 is picked up while the optical system and an image pickup element are shifted in relative position by a driving means 48. An image signal read out of a line sensor 32 is passed through a BPF 37 to extract its specific frequency component, whose output amplitude is detected by a detector 38 and divided by a divider 41 by the counted value of a counter 35. The image signal which is corrected with the storage time is added by an integration circuit 42 to output a focus signal, which is filtered by a filtering circuit 45. A microprocessor 46 detects and calculates a focusing state and a focusing position by using plural values of the focus signal and inputs them to the circuit 48 to perform focusing adjustment. Consequently, the S/N is improved to suppress the maximal point and the fast, high-accuracy focus adjustment is performed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an automatic focusing device for a camera that performs focus adjustment using a so-called hill climbing method, and a salt-containing position detection method.

[Conventional technology]

As a conventional focusing method, a predetermined frequency component is extracted from an image signal obtained from an image sensor, and the photographing optical system is moved to a position where the output amplitude of the extracted frequency component is maximized. There is a so-called mountain-climbing method for adjusting focus.

An automatic focusing device for television cameras that uses this type of mountain climbing method is described in, for example, the NHK Technical Report. Showa 40. It is explained in Volume 17, Issue 1, Series No. 86 (P21-P37).

An automatic focusing device to which the hill-climbing method disclosed in the above-mentioned document is applied will be described with reference to FIGS. 34 and 35.

FIG. 34 is a diagram showing the structure of the automatic focusing device. In this automatic focusing device, a subject image formed by a photographing optical system 1 is focused on an image sensor 2, photoelectrically converted, and output as a video signal. The video signal output from the image sensor 2 is amplified by a preamplifier 3 and then input to a bandpass filter (hereinafter referred to as rBPFJ) 4, where predetermined frequency components are extracted. Then, the output amplitude of this extracted frequency component is detected by the detector 5 and the peak detection circuit 6. FIG. 35 shows the output amplitude of the frequency component of the video signal extracted by the BPF 4.

As shown in the figure, the output amplitude of the video signal has a maximum value at the in-focus position. The output amplitude exhibiting such characteristics is held for each field of the video signal by the sample and hold circuit 7, and outputted to the one-field delay circuit 8 and the comparison circuit 9. The comparison circuit 9 compares the previously held value sent from the 1-field delay circuit 8 with the currently held value sent from the sample hold circuit 7, and determines the direction in which the output amplitude increases based on the comparison result. The motor drive path 10 is driven and controlled so that the imaging optical system moves to . Then, the photographing optical system 1 is moved to the focusing position by the motor 11.

For example, in the focusing device, if the previously held output amplitude value is at level A shown in FIG. 35 and the currently held output amplitude value is at level B, the comparison circuit 9 determines that A<B. Then, the motor drive circuit 10 is controlled to move the photographing optical system 1 closer to the in-focus position so as to maintain the moving direction of the photographing optical system 1 as it is. Then, when the output amplitude value sent from the sample hold circuit 7 to the comparison circuit 9 reaches the E level, the comparison circuit 9 determines that D>E, that is, the photographing optical system 1 has passed the in-focus position. Based on this judgment, the driving direction of the motor 11 is reversed, and the photographing optical system 1 is moved toward the in-focus position. By repeating such operations, a steady state is reached while vibrating in small increments near the in-focus position, and focus adjustment is performed.

Further, as another focusing method, a so-called phase correlation method is known in which focusing is adjusted using two light beams passing through different pupil positions of the photographing optical system. This phase correlation method is published, for example, in Minolta Techno Report (1986).

FIG. 36 is a diagram showing the structure of an automatic focusing device to which the above phase correlation method is applied. Reference numeral 21 shown in the figure is an imaging lens, and after the subject image captured by this imaging lens 21 is once formed on an imaging plane F, the separator lenses 22a and 2
2b, the image is re-imaged onto the imaging elements 23a and 23b. Note that 24a and 24b are aperture masks, which have a function of allowing only light beams of a specific F number to pass through. The image signals obtained by photoelectric conversion by the image sensors 23a and 23b are amplified by preamplifiers 25a and 25b, and then
/D converter 26a. , 26b performs A/'D conversion,
The digitized image signal is input to the microprocessor 27. The microprocessor 27 includes an image sensor 23a,
The inter-image distance i! of the subject images respectively visualized in 23b! i
ld is calculated, a control signal is output to the motor drive circuit 28, and the motor 29 is driven to perform focus adjustment. here,
Since the inter-image distance is small when the front focus is on, and large when the rear focus is on, the microprocessor 27 uses correlation calculation to calculate the inter-image distance and detects the defocus amount and focal direction.

[Problem to be solved by the invention]

However, in the automatic focusing device that uses the hill-climbing method shown in Fig. 34, the output amplitude values of the video signals are compared by 17 field times, so one focusing operation takes one field's worth of time. Therefore, it has the disadvantage that the focusing time is not long when used as an automatic focusing device for a camera.

Furthermore, the output amplitude of the frequency component extracted from the BPF4 does not necessarily draw a smooth curve due to electrical noise, camera shake, sudden changes in the subject (such as when an object passes momentarily), etc. As shown in FIG. 37, there may be a plurality of local maximum points P1 to P4. If hill-climbing focusing adjustment is performed using such an output amplitude curve, each of the local maximum points P1 to P4 will be determined to be the in-focus position, resulting in a significant decrease in focusing accuracy.

Furthermore, in the case of an automatic focusing device using a phase correlation method, since the aperture masks 24a and 24b cut off the amount of incident light in a predetermined area, the amount of incident light is significantly reduced. Therefore, in the case of a dark subject, it is necessary to lengthen the charge accumulation time of the image sensors 23a and 23b in order to photograph with proper exposure, and the photographing optical system cannot be driven while charge is being accumulated. Focusing takes a long time.

Further, in the phase correlation calculation, when the subject has a periodic pattern or the like, it is not possible to obtain an accurate distance between images, resulting in a decrease in focusing accuracy.

Furthermore, Japanese Patent Application Laid-open No. 61-32669 describes an improvement of the mountain-climbing method in which the signal is improved by adding the signals of the preceding and following horizontal scanning lines for an image with a low signal-to-noise ratio, such as a dark subject. Although a method for improving the S/N ratio has been disclosed, this method may conversely reduce the S/N ratio of the signal if the correlation between adjacent horizontal scanning lines is different.

In addition, as a countermeasure for the case where the output amplitude of BPF4 has multiple maximum points P] to P4, Japanese Patent Application Laid-Open No. 58-215176 discloses that changes in the subject are detected by changes in the aperture value, and the aperture value suddenly changes. In this case, a method is disclosed in which focusing adjustment is stopped for a certain period of time and then restarted. Furthermore, Japanese Patent Application Laid-open No. 107312 discloses that changes in an evaluation function sampled at regular intervals are monitored, and changes are n (n≧
2) Means for detecting a change in the distance between the lens and the subject when the same continuous occurrence occurs at a predetermined level is disclosed. However, the methods disclosed in both publications
There was a problem in that it took a long time to focus.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an automatic focusing device that can perform focus adjustment at high speed and with high precision without being affected by the condition of a subject.

Another object of the present invention is to provide a focus position detection method that can be applied to the above-mentioned automatic focusing device to enable extremely accurate focus adjustment.

[Means to solve the problem]

The present invention has taken the following measures to solve the above problems and achieve the objects. That is, an image sensor that captures an image formed by a photographic optical system, a drive means for changing the relative position of the image sensor and the photographic optical system in the optical axis direction, and a drive means that changes the relative position. means for capturing an image at a predetermined time interval while changing the charge and reading out the charge accumulated in the image sensor as an image signal; and a frequency extraction means for extracting a frequency component in a predetermined band from the image signal read by the readout means. and a focus signal generation means for detecting the output amplitude of the frequency component extracted by the frequency extraction means and adding the detected output amplitude values to generate a focus signal, and a focus signal generation means output from the focus signal generation means. A filtering means that performs filtering processing such as cutting high frequency components included in the focus signal, and a plurality of values of the focus signal output from the filtering means are used to perform calculations for detecting the focus state and focus position, and calculate the calculation results. and a control means for outputting a control signal based on the above to the driving means.

In addition, in order to achieve the above object, a non-destructively readable image sensor that captures an image formed by a photographing optical system, and a relative α position between this image sensor and the photographing optical system are changed in the optical axis direction. a drive means for capturing images at predetermined time intervals while changing the relative position by the drive means, and non-destructively reading out the electric charge accumulated in the image sensor as an image signal; a frequency extracting means for extracting frequency components in a predetermined band from the image signal, detecting the output amplitude of the frequency component extracted by the frequency extracting means, and adding the detected output amplitude values to obtain a focal signal. The difference between the focus signal generation means and the focus signal output from the focus signal generation means is calculated, and a plurality of values of the focus signal obtained by this difference calculation are used to determine the focus state and the focus position. The apparatus is configured to include a control means for performing a detection calculation and outputting a control signal based on the result of the calculation to the driving means.

In order to achieve the above object, the camera also includes a measuring light means for determining the brightness of the image formed by the photographing optical system and setting the readout time interval in the readout means according to the brightness.

[Effect]

According to the automatic focusing device of the present invention, an image formed by the photographing optical system is captured by the image sensor without moving the relative position of the photographing optical system and the image sensor by the driving means. This captured image is a picture @! No. 13 is read out, and frequency components of a specific band are extracted by a frequency extraction means. The output amplitude of this extracted frequency component is detected by a focus signal generation means, and a focus signal generated by adding a plurality of image signals is output. This focus signal is filtered by a filtering means. As a result, when the S/N ratio of the focus signal is improved, the maximum point of the focus signal is suppressed.The control means uses a plurality of values of the filtered focus signal to and a focus position detection calculation is performed.

When the focus position is calculated by this calculation, a control signal is output to the driving means, and focus adjustment is performed.

Further, according to the automatic focusing device of the present invention, an image signal is read out from an image sensor that can be read out non-destructively, and the frequency extraction means,
The focus signal is converted into a focus signal by a focus signal generating means. This focus signal is subjected to a difference calculation using a difference in position when the relative position between the photographing optical system and the image sensor is changed by the control means, and the focus signal is subjected to filtering processing. Then, a focus position is calculated using the filtered focus signal, and a control signal is output to the driving means based on the detected focus position to perform focus adjustment.

In addition, by providing a photometric means, it is possible to estimate the amount of charge accumulated in the image sensor, and since the readout time interval is set according to the brightness of the image, it is possible to effectively prevent the amount of charge accumulated in the image sensor from becoming saturated. Become.

〔Example〕

FIG. 1 is a diagram showing the configuration of an automatic focusing device according to a first embodiment of the present invention. 31 shown in the diagram is a photographing optical system,
An image formed by this photographing optical system 31 is formed on a line sensor 32 made of a COD or the like. This line sensor 32 transfers charges accumulated at predetermined time intervals to an image f5.
It is output to the pre-amplifier 33 as the number SO, and the stored vX
A peak signal PE indicating the amount of charge is output to the drive circuit 34.

A specific configuration of the line sensor 32 is shown in Figure 2.

This line sensor 32 includes n photosensors 51-1
51-n are arranged in a row, and the photosensors 51-1
A CCD shift register 53 is arranged on one side of the columns 51-n through a transfer gate 52. This shift register 53 is connected to a terminal 55 via an output amplifier 54.
The image signal is output from this terminal 55. In addition, each of the photosensors 51-1 to 51 arranged in a row
-n is connected to the vinyl detector 57 via buffers 56-1 to 56-n. This peak detection section 57 selects a peak value from the signals sent from each of the photosensors 51-1 to 51-n, and outputs a peak signal PE from a terminal 58.

Note that the terminal 5 to which the transfer 7 gate 52 is connected
A drive pulse φ1, which will be described later, is inputted to the input signal.

The drive circuit 34 is a drive circuit for the line sensor 32, and outputs a clock pulse φS and a drive pulse φd to the line sensor 32, and also outputs a reset signal to the counter 35 and a peak signal PE when the peak signal PE reaches the saturation amount. Issues a count stop signal. The counter 35 counts the clock pulses φ0 sent from the oscillator 36, and outputs the count number N to a divider to be described later. On the other hand, the image signal amplified by the preamble 33 is manually input to the BPF 37. BPF 37 extracts a predetermined frequency component of the image signal and outputs it to detector 38 . The detector 38 detects the output amplitude of the extracted frequency component. The output of this detector 38 is A/D converted by an A/D converter 39 and then output to the divider 41. The divider 41 is the A/D converter 39
The digital data inputted from the counter 35 is divided by the count number N sent from the counter 35, and the accumulated data I1 is calculated for the image signal.
．． Corrects values between ¥. 42 is an integral channel, which is composed of an adder 43 and a latch 44, and has a focus signal f (
X) is output to the filtering circuit 45.

The filtering circuit 45 performs filtering processing to be described later and outputs a filtered focus signal (hereinafter referred to as "filtering signal") g(x) to the microprocessor 46. This microprocessor 46 stores the filtering signal g (x) sent from the filtering circuit 45 in the memory 47, and also stores the filtering signal g (x) stored in this memory 47.
) to perform arithmetic processing to determine the in-focus state and in-focus position. A motor drive circuit 48 drives the pulse motor 49 based on a control signal sent from the microprocessor 46.

Next, characteristic signal processing in this embodiment will be explained.

In this embodiment, in order to prevent the signal S/N ratio from deteriorating due to saturation in the accumulation of the line sensor 32, when the peak signal PE output from the line sensor 32 is saturated, the line sensor 32 is stopped and read out, the image signal value is converted from the time until saturation is reached, and the supplementary W. is added. This will be explained in detail with reference to FIG.

The figure shows the relationship between the amount of charge accumulated and the accumulation time in the line sensor 32, with the horizontal axis representing the rfs 6:j accumulation time and the vertical axis representing the amount of accumulation 1jS +6f. ts is a unit accumulation time and is equal to the detection time interval of the focus signal.

e, indicates the saturation amount of accumulated charge. Here, three types of images i, (r), i2(r), is
The amount of accumulated charge at the maximum signal position of (r) is the straight line P I +
Let us consider a case where P 2 *P changes. Note that "" indicates the position of the line sensor 32.

Both images i. (r) is the case where the peak value e,<e5 and saturation has not occurred, and the amount of accumulated charge in unit accumulation time 1 is at all line sensor positions ``it(
r)<es. Image i-(r). i- (
r) is a case where at least the signal peak value is saturated, and charge accumulation is stopped at peak value saturation time t2+t, and reading is performed. Then, this is converted into a value at unit accumulation time t, using the following equation.

i' 2 (r)-i2 (r)Xts/t21' 3
(r)-is (r)Xt3/t, or simply 1' + (r) - 1 + (r)/tsi'
2 (r)-i2(r) /t2i' 3 (r)
- is (r)/t3. That is, by dividing by the accumulation time, differences in signal output due to different accumulation times are corrected.

Next, filtering processing in the filtering circuit 45 will be explained. If h (x) is a filter function and "*" is a convolution symbol, then the filtered signal g (x) is g (x) = f (X) *
It can be expressed as h (x) − (1.).

Examples of the filter function h(X) include a rect function, a spline function, and a sine function. Here, rec
t function is rect(x/a), spline function is spl
ine(x/a) and is defined as follows, for example.

rect(x/a): When X1≦a/2 rect(x/a)-1XI>a
/2 when rect (x/a)=0...(2) spline(x/a): when X1≦a/2, spline(x/a)-X" - 2 X +
1a/2 < l X When l≦a, spline (x./a) = − X3+5 X −
When 8 Moreover, the spectra of both functions are shown in FIGS. 5(a) and 5(b), respectively. As is clear from the figure, it has the characteristics of a low-pass filter, and by performing filtering processing using both functions, high frequency components such as local maximum points are suppressed.

By comparing FIG. 5(a) and FIG. 5(b), it can be seen that the spline function is closer to a low-pass filter. For example, as shown in Figures 5(a) and (b), if the Nyquist frequency is set to U 3 x (1/a), the sampling pitch ΔX of the filtering function will be (a/2) x (1/a). 3) -a/6, and filtering processing by convolution can be performed using 6 impulse response signals for the rect function and 12 impulse response signals for the spline function.

As a circuit that actually performs such filtering processing, a circuit having the structure shown in FIG. 6(a) can be considered. This is done by delaying the focus signal f (X) one after another by delay elements T such as a shift register, and inputting rec to the output of each delay element.
Coefficients W1, W2, as t functions or spline functions
. . . After multiplication by wm, an adder adds the result and outputs it as a filtered signal g (x).

Moreover, as shown in FIG. 6(b), the focus signal f (x)
may be multiplied by coefficients w1, w2, .about.wm, respectively, delayed by delay elements T, and sequentially added to output the filtered signal g(x).

Furthermore, when using the rect function as a filtering function, the coefficients wl-w2-...-w m = 1
Therefore, it is a simple addition of signals, and the circuit structure shown in FIG. 6(C) can be obtained.

By performing such filtering processing, as shown in Fig. 7, the focal signal f(x) shown by the line with maximum points at multiple locations is suppressed, and the multiple maximum points shown by the broken lines in the figure are suppressed. It is converted into a filtered signal g (x). Note that ΔX indicates the detection interval of the focus signal.

Next, the arithmetic processing for determining the in-focus state and in-focus position will be explained with reference to FIGS. 8(a) and 8(b). 8th
Figure (a) is a diagram showing a filtering signal, in which the vertical axis shows the output level g (X) of the filtering signal, and the horizontal axis shows the position X of the photographing optical system 31. now,
Distance i in the horizontal axis direction! Let two points g (XI) and g (xi-ff) on the filtered signal separated by tIN be V1 and V2, respectively.

Note that g (XI) is the most recently calculated filtering signal value. First, the magnitudes of v1 and v2 are compared, and if Vl>V2, the driving direction of the line sensor 32 is maintained as it is. Then, V1 and v2 are moved in the horizontal axis direction, and a position where v1 passes the in-focus position and V1<V2 is detected. Then, Vl<V2 across the focus position
A pair of two points (hereinafter referred to as Va and Vb) and Vl
>V2 (hereinafter referred to as Vc and Vd) are set. The four points Va, Vb, Vc, set in this way,
The positional relationship of Vd is shown in FIG. 8(b). 4 points Va, Vb
, Vc, and Vd, line segment VaVc and line segment V
Find a straight line m that crosses bVd and is parallel to the X axis. Then, the point of contact between the straight line m and the line segment VaVc is set as C1,
The contact point between the straight line m and the line segment VbVd is set as C2, and the contact point CI
, the distance L2 in the X-axis direction between points Va, and the contact point C2 and point Vb
Directly 1jl so that the distance LI in the X-axis direction between
Set the position of m in the Y-axis direction. And line segment C1C2
Find the midpoint M of , and set the X coordinate of this M as the focus position α.

This α can be expressed by the following formula.

α-XVa+N/2+ΔX (Vb-Va) / (V
c-Vd+Vb-Va)... (1). The in-focus position can be detected by processing the above equation (1) using a microprocessor.

As a result of experiments, when focusing was performed using the above-mentioned focusing position detection method, the detection error was zero at the quadratic function level. Moreover, XVa in the above equation (1) indicates the X coordinate of Va. The distance AtlN is set to an appropriate distance depending on the defocus amount, image, zoom position, and F number. For example, if this distance is set large, the difference in filtering signals g (x) will become large, making it easier to obtain an S/N ratio during calculation.

Next, the operation of the automatic focusing device that performs the configuration and signal processing described above will be explained. The photographing optical system 31 is driven at high speed, for example, in a direction approaching the common focal distance (distance where photographing is frequently performed). At this time, in the line sensor 32, the transfer gate 52 causes the photosensors 51-1 to 51 to
-n is reset and starts accumulating charges. At the same time, the counter 35 is also reset, and a clock pulse φ is generated. counting begins. If the peak value signal PE does not reach the saturation amount even after the predetermined readout interval of unit accumulation 1i has elapsed, the counter 35 stops counting at a point where t has elapsed, and the transfer gate The image signal SO is output to the preamplifier 33 via 54. At the same time, the counter 35 is reset and the next charge accumulation is started.

Further, when the peak value signal PE shows the saturation value before the elapse of time t5, the image signal SO is outputted at that point, and the counter 35 stops counting. Then, after time t5 has elapsed, the photosensors 51-1 to 51-n and the counter 35 are reset to start the next charge accumulation. The image signal read from the line sensor 32 is sent to the preamplifier 33.
After being amplified by the BPF 37, a predetermined frequency component is extracted, and the output furisode of this extracted frequency component is sent to the detector 3.
Detected at 8. Then, the A/D converter 39
The converted and digitized signal is divided by the count number sent from the counter 35 in a divider 41. That is, as described above, the image signal is corrected to a value corresponding to the unit accumulation time. The image signals corrected according to the accumulation time are sequentially input to the integrating circuit 42. Here, n image signals are added for the purpose of improving the S/N ratio of the focus signal f (x). In this way, different values of the position by Δχ are obtained every ts time, and the focus signal f (x) shown by the solid line in FIG. 9 is detected. This focus signal f (X) is input to the filtering circuit 45 and subjected to the filtering process described above. This filtering process improves the S/N ratio of the focus signal f (x) and suppresses the maximum point, resulting in a filtered signal g (X) shown by the broken line in FIG. 9. The filtered signal g (x) thus obtained is input to the microprocessor 46 and sequentially stored in the memory 47 . Then, using the filtering signal value stored in the memory 47, the focus position α is determined by the method described above. A control signal based on the determined focus position α is output to the motor drive circuit 48, and the pulse motor 49 is driven. As a result, the photographing optical system 31 moves to the focus position and focus adjustment is performed.

In this way, according to the present embodiment, the image signal read from the line sensor 32 is divided by the count number corresponding to the accumulation time, so that the image signal can be corrected to the value for the unit accumulation time. Even in the case of overexposure, imaging can be performed without damaging the image signal.

In addition, the output amplitude of the frequency component extracted by BPF37 is
Since a plurality of signals are added by the integrating circuit 42, the S/N ratio of the output focus signal can be improved.

Further, since the filtering circuit 45 convolves the focus signal f (x) with the spline function or the rect function and performs the filtering process, it is possible to ensure the S/N ratio of the focus signal f (x). Specifically, in the filtering process using the rect function shown in FIG. 4(a), the filtered signal g (X ), and the signal component is approximately multiplied by the number of additions, while the noise component is multiplied by the square root of the number of additions, so the S/N ratio is improved by approximately Jτ times.Also, considering in the frequency domain , the signal component is proportional to the bandwidth,
Since the noise component is proportional to the square root of the bandwidth, the S/N ratio is improved approximately five times even when a spline function is used. Therefore, since the S/N ratio can be ensured, the electric current Gj can be read out at short intervals, and the time required for focusing can be shortened. Moreover, since the S/N ratio can be secured, the detection interval ΔX of the focus signal can be made small.
Since the maximum point of the focus signal f (x) can be suppressed, focusing accuracy can be improved.

Furthermore, since the focus position is detected using the focus position detection method described above, the focus position can be detected with extremely high accuracy.

In the first embodiment, one BPF 37 is provided to obtain the focus signal, but a plurality of BPFs for light at the center frequency may be switched and used. With such a structure, the following inconveniences can be eliminated. That is, when the center frequency of the BPF is low, the obtained focus signal has a broad shape as shown in A shown in FIG. I can't ask for it. Furthermore, when the center frequency of the BPF is high, the obtained focus signal has a sharply protruding shape at the focus position as shown by B in Figure 10, which is suitable for detecting the focus position with high precision. However, the detection range of the position of the photographic optical system is narrow. Therefore, with the above configuration, when the defocus amount is large, the BPF with a low center frequency is switched to, and when the photographing optical system moves near the in-focus position, the BPF with a high center frequency is switched.
Switch to Also, ii! It may be switched depending on the ii image, zoom position, F number, etc.

In the above embodiment, the BPF 37. 38 detectors, 3 A/D converters,
Although the divider 41 and the integrating circuit 42 have the configuration shown in FIG. 1, they may have the configuration shown in FIG. 11(a). That is, after the image signal output from the briambu 33 is A/D converted by the A/D converter 61, it is divided by the count number N from the counter in the divider 62. Then, a predetermined frequency component is extracted from this image signal by a digital bandpass filtering circuit 63, this frequency component is further input to a detection circuit 64, and the output amplitude detected by this detection circuit 64 is integrated by an integration circuit 65. composition. With this configuration, various signal processing is performed after the image signal output from the preamplifier 33 is A/D converted, so that the dynamic range of the signal can be effectively used. Moreover, as shown in FIG. 11(b), BPF71. Detector 72, integrating circuit 73, sample hold circuit 74. A/
D converter 75. The divider 76 may also be a groove. In addition,
As the integrating circuit 73, it is also possible to use a low-pass filter. By configuring in this way, A/
It is sufficient to perform D conversion at each focus signal detection time interval, and A
The operation clock for /D conversion can be slowed down, and the configuration can be simplified and manufacturing can be facilitated.

Furthermore, the output of the preamplifier 33 may be A/D converted and all subsequent signal processing may be performed within the microprocessor.

Next, a second embodiment of the present invention will be described.

FIG. 12 is a diagram showing the configuration of an automatic focusing device according to a second embodiment. This automatic focusing device uses S I as an image sensor.
T (Static InducLIon Tran
slsLor) ,AMI(^ppllf'led M
OS Intelligent Imager),C
M D (Charge Modulation D
It is equipped with an image sensor (hereinafter referred to as a "non-destructive element") 71 that can be read out non-destructively, such as VLCE, MOS, etc., and after converting the image signal read out from the non-destructive element 71 into a focus signal, The feature is that a difference calculation of focus signals is performed and filtering processing is performed. Note that the same parts as in the first embodiment shown in FIG. 1 are given the same reference numerals.

The image signal read out from the nondestructive element 71 is amplified by the amplifier 33, and then the BPF 37 and the detector 38 detect the output signal of a predetermined frequency component, and this detected value is sent to the A/D converter 39.
It is manually input to the integrating circuit 42 via. And the integrating circuit 4
2, the focus signal f N (X) is output to the microprocessor 73 at predetermined intervals. This microprocessor 7
The focus signal f N (X) inputted into the memory 74 is stored in the memory 74 . The microprocessor 73 calculates the difference using the difference in the position of the photographing optical system 31 on the optical axis using the focus lens 3 (X) stored in the memory 74, and generates a filtered signal g ( Find X). Then, a compensating calculation using the filtering signal g (X) is performed to detect the in-focus position, and a control signal is output to the motor drive circuit 48 based on the detection result. The motor drive circuit 48 drives the pulse motor to move the position of the photographing optical system 31 and adjust the focus.

FIG. 13 is a diagram showing the structure of the non-destructive element 71.

Reference numerals 81-1 to 81-n shown in the figure are nondestructive readout photosensors, and readout switches 82-1 to 82-n are connected to the photosensors 81, respectively. The readout switches 82-1 to 82-n are connected to a scanning circuit 83, and the image signal driven and read out by the scanning circuit 83 is output as an image signal from a terminal 85 via an output amplifier 84. Note that 86 is a reset switch.

Each of the nondestructive photosensors 81-1 to 81-n is connected to a peak detection section 88 via a buffer 87-1 to 87-n, and a peak value is detected from the peak detection section 88 and sent to a terminal 89. The peak value is output as No. 18 PE.

Next, a filtering process based on a difference calculation using a difference in the position of the photographing optical system 31 on the optical axis of the focus signal f (X) will be described.

t1 while moving the photographing optical system 31 at a constant speed in the optical axis direction.
When charge is accumulated for a certain amount of time, the image signal obtained at that time becomes an integral value of the image signal within the movement range of the photographing optical system 31. Further, the focus signal obtained from the image signal is also an integral value of the focus signal within the movement range of the photographing optical system 31.

In other words, the focus signal f N (XI) when the photographing optical system 31 moves from the position XO to the position X1 at time t is f N (!+) - f''r(x)dx
--- (4)x0, which is equal to the area of the shaded area shown in FIG. Then, f N (X) can be expressed as shown in Figure 15, and f (X) is obtained by differentiating f N (X).

Further, the difference in the focus signal f N (X) (the difference in the optical axis direction of the photographing optical system 31 is assumed to be Px) is − f (X) * reel (−) Px (5), and the difference in the photographing optical system 31 is as follows. This means that the focus signal f (X) at each position when the system 31 moves in the optical axis direction is filtered using the rec t function.

As described above, when the non-destructive element 71 is used, the filtering process is replaced with a difference calculation.

Note that, in reality, since the non-destructive element 71 becomes saturated, the accumulated charge of the non-destructive element 71 is reset at the point IJj where the peak value signal PE is saturated. Therefore, the focus signal f N (X) shown in FIG. 15 changes as shown by the solid line in FIG. 16. This figure shows that the element is reset when the photographing optical system 31 is at the positions Xa and Xb in the optical axis direction.

And if Xa<X<Xb, f N (x)=f' N (X)+f N (xa)
For xb<x, f N (x)−f N (x)”f N
It is obtained as (xa)+f N (xb).

Further, the focus position detection in this embodiment is obtained by interpolation calculation using a plurality of points near the maximum value of the focus signal.

This will be explained below with reference to FIGS. 17(a) and 17(b).

FIG. 17(a) is a diagram showing the focus signal, where the horizontal axis shows the position of the photographing optical system 3 moving in the optical axis direction, and the vertical axis shows the signal level of the focus signal g (X). . At point PO, which is the maximum value of the focus signal, and at three points before and after this point, p, , p2, the focus signal values are g (Xl1).
Suppose that g (xn+-1) and g(XII+1).

Here, if g (xm-1) < g (xn+l), the intersection point Px of a straight line connecting points Pa and P, and a straight line whose slope is opposite in polarity to this straight line and which passes through point P2. The coordinate α on the horizontal axis is the focal position. This coordinate α is obtained from the following equation based on the geometric relationship.

a −Xm+ ΔX /2 [ 1 g (Xm+1)
- g (Xm-1)1/ l g (Xm)- g
(Xs-1)l ] − (II) Also, g (x
When m-1) > g(xw+1), 1 is calculated by the following formula.

a -Xm- ΔX12[1 g (Xm-1)- g
(Xm+1))/(g (Xm)-g (Xm
+1))] - (ml) Next, the operation of this embodiment will be explained.

The non-destructive element 71 is reset and charge accumulation is started. Then, every time 1, the readout switches 82-1 to 82-
n is scanned to read accumulated charges from the non-destructive photosensors 81-1 to 81-n and output them as image signals. The read image signal is amplified by the preamplifier 33, and the BPF
At 37, predetermined frequency components are extracted. After this frequency component is detected by a detector 38 and A/D converted, n image signals are added together by an integrating circuit 42, and this added signal is output as a focus signal f N (x). and,
When the value of the peak signal PH output from the non-destructive element 71 reaches the saturation point, a reset signal is output from the drive circuit 72, and the non-destructive element 71 is reset. This focusing operation produces a focus signal f N (x
) is obtained. This focus signal f s (X) is the memory 7
4 is stored.

Further, times ta and tb shown in FIG. 18 are the times when the saturation point is reached, and the positions Xa and Xb of the photographing optical system 31 at these times are also stored in the memory 74. The focus signal f N (X) stored in the memory 74 is a filtered signal g (
After performing the difference calculation to obtain x), unnecessary items are deleted.

The difference calculation for obtaining the filtered focus signal g(x) is performed by calculating the difference in position on the optical axis of the photographing optical system 31 by d.
Then, g (X) − f N (X+d/2) f N
It can be expressed as (X-d/2) − (8).

Here f N (x) is When X<Xa, f N (X) ...(9) It is.

The filtering signal g(X) obtained in this way is sorted as Vi, V2, and V3 in order of newest, and the microprocessor 73 determines the focus state from these three values and controls and drives the pulse motor 49. In other words, Vl > V2
>V3, the focus signal increases, so the photographing optical system 31 is driven as is; when Vl<V2<V3, the filtering signal decreases, so the driving direction is reversed. And Vl<V2, V3<v2
In this case, it is determined that V2 is the maximum value of the filtering f self-sign, and the focusing (vertical calculation) is performed using an interpolation calculation.
Xa) − V 2 , g (Xw+I) − V
1, this is done by substituting it into the above equation (If) (III).

The focus position obtained in this way and the pulse motor 49
The driving amount is determined from the current position of the photographing optical system 31 obtained from the number of pulses during driving, and the photographing optical system 31 is driven to the in-focus position.

According to this full-length embodiment, it is possible to perform focus adjustment at high speed and with high precision, as in the first embodiment. Moreover, since the image sensor 7 that can be read out non-destructively is used, the focus signal g (x) filtered by the rect function can be obtained by calculating the difference between the focus signals f N (x). , the filtering circuit 45 used in the first embodiment can be omitted, and the device can be made smaller.

In addition, in the second embodiment, each value of the filtering signal V1, V2, V3 (however, Vl<V2, V3<
V2), the detection interval of the focus signal can be reduced by reversing the driving direction of the photographing optical system 31 and slowing down the driving speed before adjusting the focus, without immediately performing the compensation calculation. ΔX can be shortened, and accuracy can be further improved.

Next, a third embodiment of the present invention will be described.

FIG. 19 shows the configuration of an automatic focusing device according to a third embodiment. Note that the same parts as in the first embodiment or the second length embodiment are given the same reference numerals, and detailed explanations will be omitted.

This embodiment is an example in which the brightness of an image formed by the photographing optical system 31 is measured and the readout time interval from the non-destructive element 71 is set. The non-destructive element 71 is driven by a read clock φC transmitted from the drive circuit 90. The peak signal PE of the accumulated fa charge of the non-destructive element 71 is detected by the comparator 91.
The input signal is manually input to the D flip-flop circuit 92 via the D flip-flop circuit 92. The threshold value Vref of the comparator 91 is set to a value slightly smaller than the peak value (saturation level) of the accumulated charge of the non-destructive element 71. The comparator 91 outputs H(
high level) output PK to the D flip-flop circuit 92.
Output to. The D flip-flop circuit 92 is a comparator 9
PK from 1 is used as the D input, and read clock φC from the drive circuit 90 is used as the CLK input. Then, when the D input is H and there is an input to the CLK input, the Q output becomes H, and when there is an input to the CLK input again, the Q output becomes L.
(low level). The Q output of the D flip-flop circuit 92 is output to the non-destructive element 71 as a reset signal for the non-destructive element 71 when it is H. Note that the drive circuit 90
The timing of the read clock φC output from the microprocessor 93 is set by the microprocessor 93. Non-destructive element 7
The image signal read from 1 is input to an amplifier 33.

A BPF 37 and a pre-photometering circuit 94 are connected in parallel to the output terminal of the amplifier 33. The optical circuit 94 measures the brightness of the image formed on the non-destructive element 71 from the manually inputted image signal, and outputs the measurement result to the microprocessor 93. Note that from the image signal output to the BPF 37, the detector 38, A/D converter 39, and integration circuit 43
The configuration for generating the focus signal f N (x) via is the same as in the second embodiment.

Next, the operation of this embodiment will be explained.

The brightness of the image is measured by the blur photometry circuit 94.

This photometry is performed from the start of imaging to the D flip-flop circuit 92.
TV is performed until the D human power becomes H.

That is, the peak value PE of the nondestructive element 71 is equal to the threshold value V
Continue until it exceeds ref.

At this time, if the time tv is shorter than the unit accumulation time IS described in the first embodiment, when the subject is sufficiently bright and the subject is exposed for the unit accumulation time ts, the accumulated fa electricity Gj of the non-destructive element 71 will be reduced. This is a case of saturation. Hereinafter, this will be referred to as saturation mode.

Therefore, in this embodiment, the microprocessor 93 uses the photometric data sent from the pre-photometric circuit 94 to
Set a unit accumulation time ts such that ts < tv.

Then, the non-destructive element 71 is reset every unit accumulation time t, and the image signal SO is read out.

If tv is longer than the unit accumulation time ts, the subject is dark and the accumulated charge in the non-destructive element 71 is not saturated with exposure for the unit accumulation time t. Hereinafter, this will be referred to as non-saturated mode.

FIGS. 20(a) to 20(d) are diagrams showing the reset operation in the non-saturation mode. The figure (a) shows the non-destructive element 71. ! ! The output level of the peak signal PE indicating the load accumulation state is shown. FIG. 4B is a timing waveform diagram of the peak signal PE input to the D input of the D flip-flop circuit 92, and FIG. FIG. 3 is a timing waveform diagram of a read clock φC which is manually operated as CLK. Same figure (d)
shows the timing waveform of the reset signal R serving as the Q output of the D flip-flop circuit 92.

When imaging by the non-destructive element 71 is started, an image signal is read out every unit storage time ts. Charge is accumulated in the nondestructive element 71, and when the output level of the peak signal PH exceeds the threshold value Vref at time S1, the D input becomes H. The read clock φC output at the next timing after the D input becomes H causes the Q output to become H, and the non-destructive element 71 is reset. Then, the Q output becomes L by the read clock φC output at the next timing. Then, when the output level of the peak signal PE exceeds the threshold value Vref again at time S2, the reset operation is similarly started.

As described above, according to this embodiment, the brightness of the image is measured by the pre-photometering circuit 94, and in the saturation mode, the brightness of the image is measured at t5.
< tv (set the 11th accumulation time ts, i
The non-destructive element 71 is reset every 11 accumulation times ts and the image signal SO is read out.

In addition, in the non-saturation mode, the unit accumulation time which is the output timing of the read clock φC is set based on the luminance data, and the saturation level is set for the peak signal PH indicating the charge accumulation state of the non-destructive element 71. The threshold value Vref is set to a level slightly lower than the threshold value, and when the output level of the peak signal PE exceeds the threshold value, the non-destructive element 71 is reset by the read clock φC output at the next timing. I made it. therefore,
Regardless of the brightness of the subject, saturation of the non-destructive element 71 can be reliably prevented, deterioration of the image signal can be prevented, and highly accurate focus adjustment can be performed.

In the third embodiment, the focus signal fN(X) is not filtered in the saturation mode, but the microprocessor 93 may perform the filtering process.

Next, a fourth embodiment of the present invention will be described.

FIG. 21 is a diagram showing the structure of an automatic focusing device according to a fourth embodiment. Note that parts having the same functions as those in the first to third embodiments are designated by the same reference numerals, and detailed description thereof will be omitted.

This embodiment is an example in which an area sensor is used to focus on an arbitrary area of an image: section A.

100 shown in the figure is an area sensor, MOS, SI
It consists of an x-y addressing type image sensor such as T or CMD. This area sensor 100 is driven by a drive signal D output from a drive circuit 101, and image signals of a predetermined area are read out.

The blur photometry circuit 94 receives the image signal read from the area sensor 100 and measures the brightness of the image from the input image signal. The brightness data is output to microprocessor 102. The microprocessor 102 is
a function of setting an accumulation time such that a predetermined area of the area sensor 100 is not saturated based on input luminance data;
It has a function of specifying a readout area of the area sensor 100 and a function of executing a focus position detection calculation from the filtering signal g(x).

FIG. 22 is a diagram showing the structure of the area sensor 100.

In this area sensor 100, a plurality of photosensors FS are arranged in a matrix.

Each photosensor FS has a plurality of horizontal signal lines L L each having one end connected to the vertical address decoder 103.
-1 to Ll-N, each vertical address? 103S
connected via. Further, it is connected to a plurality of vertical signal lines Ll+-1 to Lll-N, each of which has one end connected to the horizontal address decoder 104, via a horizontal address switch 104S.

In the area sensor 100 configured in this way, the vertical address decoder 103 selects the predetermined horizontal signal lines L1-1 to L.
1-N, selects a predetermined vertical signal line LL+-1 to Ll1■ with the horizontal address decoder 104, and simultaneously turns on the horizontal address switch 103S and vertical address switch 104S connected to each signal line. As a result, accumulated charges are read out from the plurality of photosensors FS in the selected area as an image signal SO. Therefore,
Vertical address decoder 103, horizontal address decoder 1
By specifying the address of the readout area with the drive signal D input to 04, it is possible to read out the image signal of the (f-area) of the image.

Hereinafter, the operation of this embodiment will be explained with reference to FIG. 23. Note that the horizontal direction of the area sensor 100 shown in FIG. 22 is the X axis (X-1.2...n), and the vertical direction is Y.
The axis is (Y-1.2...n).

An example will be described in which the shaded portion in the figure is adjusted by JJA.

First, pre-photometering is performed in the shaded area.

The microprocessor 102 is located in the shaded area (CXs.
Address designation data of Ys) to (Xe-Ye)) is output to the drive circuit 101.

The drive circuit 101 outputs a drive signal D specifying an address (Ys-Ye) to the vertical address decoder 103, and outputs a drive signal D specifying an address (Ys-Ye) to the horizontal address decoder 104.
A drive signal D specifying the value (Xs - Xe) is output. Then, for example, charges accumulated during a predetermined time period 1 are sequentially read out as image signals from each photosensor FS in the shaded area. The photometric circuit 94 measures the brightness of the image from this read image signal and outputs the brightness data to the microprocessor 102. The microprocessor 102 sets a unit accumulation time ts based on the luminance data.

That is, if the peak value of the charge accumulated in the photosensor FS is not saturated and exceeds the threshold value set to 70% of the saturation value, the above accumulation time tP is used for single accumulation. The time is assumed to be 1 s.

In addition, when the peak value is saturated, the brightness determination is repeated by sequentially shortening the M Let time be t5.

In addition, if the peak value has not reached the threshold Vref, the charge accumulation time is sequentially extended and the brightness determination is repeated until the accumulation time t F at which the peak value exceeds the threshold Vref is reached.
PI' is detected, and the detected accumulation time tPPP is set as the fit accumulation time ts.

By such pre-photometering, it is possible to set an accumulation time ts of about jlj without deteriorating the image signal according to the brightness of the shaded area.

Next, the image signal SO in the shaded area is read out at every time ts.

Note that the order of reading out the image signals SO from the shaded area follows the following method.

? Axial Sequential Readout This readout method reads out data sequentially in the X-axis direction. First, Ym
Ys row ( (X s * Y s ) + (
s+ l +Yu)..., (Xe, Ys)), then read Y s+m Y 5+, row, y2+2 row, -Y
Read up to line e.

Sequential readout in Y-axis direction This readout method sequentially reads out data in the Y-axis direction.

First, the X-Xs sequence ( (Xs, Ys), (Xs, Ys+
+ ). −． (Xs, Ye)), then
-XS ■ column, X, 12th column...Read up to -, Xe column.

Sequential readout in diagonal direction This readout method reads data sequentially in diagonal direction. First, (X
s, Ys), and the following -, Y--X+ (Yi+X
s) In addition, i − S + 1, S + 2 −
Charges of pixels (X, Y) satisfying X s≦X≦Xe, Y S + l≦Y≦Ye are sequentially read out.

The above three readout methods are appropriately selected depending on the subject. For example, in the case of a subject with horizontal stripes, the sequential readout method in the Y-axis direction is used, and in the case of a subject with vertical stripes, the sequential readout method in the X-axis direction is used. This readout method is selected during blur photometry.

In this way, the charge accumulation time ts and the readout order are determined.

Next, the relationship between the read clock and charge accumulation time for the area sensor 100 will be explained with reference to FIG. 24.

The figure shows the readout clock and the pixels read out at that time (
Photosensor FS) is shown. 1~N (N-m
Xn) corresponds to the pixel position, and 1' to N' indicate the next readout position.

Here, if the time interval of the readout clock is tc, the time tN required to read out all pixels is tN-N-tc.

Here, if the unit accumulation time ts is greater than tN (the 24th
Figure (a)) can be read without any problem, but if ts<t. If this happens, we cannot deal with it.

Therefore, the same pixel is read out in two consecutive clocks (or clocks at appropriate intervals). In this way, 15<1
N (note that tN is twice that in (a) of the same figure), it is still possible to read, and the minimum time of t is the clock time interval tc ((b) and (c) of the same figure). . Furthermore, in order to read out data in a shorter time, a plurality of horizontal address decoders and vertical address decoders are provided so that reading is performed alternately. Furthermore, multi-line readout may be used to increase the readout speed, and adjacent pixels may be added to improve the SN. Note that since all pixels are read out sequentially, the integration time for each pixel is different (the integration time is the same at ts), but the focus signal is simply averaged in the optical axis direction, so no particular problems will occur. do not have.

The image signal read out in the manner described above is input to the BPF 37 as a one-dimensional signal, and thereafter processed in the same manner as in the first embodiment to perform focus adjustment.

As described above, according to the present embodiment, since the formed image is captured using the area sensor 100, it is possible to perform high-speed and highly accurate focus adjustment on an arbitrary area of the image.

In the fourth embodiment, the image signal readout method is selected by the pre-photometering circuit 94, but a dedicated circuit for determining the type of subject may be provided.

Further, if the image signals are added row by row in the case of the X-axis sequential readout method, and column by column in the case of the Y-axis sequential readout method, the SN of the signal can be further improved.

Note that as a configuration for performing such addition, for example, the 21st
The path from the BPF 37 to the A/D converter 39 shown in the figure is
As shown in FIG. 5, an A/D converter 110, an adder 111,
A configuration in which the BPF 112 and the detector 113 are connected in series can be considered.

In addition, in the fourth embodiment, the BPF 37 uses a one-dimensional signal (
Although the filter circuit is for an image signal), it may be a two-dimensional filter circuit with a built-in buffer and mask processing.

Next, an example of the brush length 5 according to the present invention will be described.

FIG. 26 is a diagram showing the structure of the full length embodiment. In addition, the first
~ Are the same parts as in the fourth embodiment given the same reference numerals? are doing. This embodiment uses C as the area sensor explained in the third embodiment.
This is an example using a CD two-dimensional image sensor.

Reference numeral 120 shown in the figure is a CCD two-dimensional imaging device, to which a transfer gate drive pulse φ1, vertical clock pulse φ, horizontal clock pulse φ■, and output gate drive pulse φOI+φo2 are inputted from a drive circuit 121. The microprocessor 122 has the same functions as the microprocessor described in the fourth embodiment, and further includes the drive circuit 12.
It has a function to set the pulse output from 1.

FIG. 27 is a diagram showing the configuration of the CCD two-dimensional image sensor 120. This CCD two-dimensional image sensor 120 has mXn pixels, and photo sensors 131, each pixel unit, are arranged in a matrix. A plurality of photosensors 131-01 to 131-O arranged in the Y-axis direction
Transfer gates 132-1 to 132-n are provided along the lines n, . . . 131-m1 to 131-mn. A transfer gate drive pulse φ7 is manually applied to each of the transfer gates 132. Vertical shift registers 133-1 to 133-m are provided along each transfer gate 132-1 to 132-n. A vertical clock pulse φ9 is manually applied to each of the vertical shift registers 133. Each transfer gate 13
2 and one end of each vertical shift register 133 are connected to a horizontal shift register 134. A horizontal clock pulse φ1I is input to the water ilZ shift register 134. The signal charge transferred to the horizontal shift register 134 is taken out via an output gate 135 formed along the horizontal shift register 134 and an output gate 136 provided at one end of the horizontal shift register 134. Note that 137, 138, and 139 are output drains. The signal charge taken out from the output drain 139 is outputted as an image signal SO via the output amplifier 140.

Further, the output gates 135 and 136 are provided with an output gate drive pulse φ. 1,φ. 2 are input respectively.

Hereinafter, as an example of the operation of this embodiment, the case where the focus is adjusted for the photosensor in the shaded area in FIG. explain.

When the charge accumulation is completed, the charges of the photo sensors 131-01 to 131-mn are transferred to the transfer gate 13.
Vertical shift register 133 via 2-1 to 132-m
-1 to 133-m. Then, it is transferred at high speed by the vertical shift register 133 to the horizontal shift register 134. The signal charge transferred to the horizontal shift register 134 is transferred to the output drain 13 via the output gate 4135.
7.

Such readout is performed for signal charges of Y-n to Y2 and all pixels in a row. Note that the vertical shift register 133
The high-speed transfer of the read signal '2l! The speed is set so as not to damage the cocoons. Then, once the signal charges of the Y-Y2 rows have been transferred to the horizontal shift register 134, this high-speed transfer operation is stopped and a normal signal read operation is started. However, signal charges other than X-X, ~X2 are at the output gate 13.
6 to an output drain 138.

After reading out the f≧ charges of the pixels in the rectangular area in this manner, the high-speed transfer operation is started again, and YY. -1~1
The signal charges from the pixels in the row are discharged to the output drain 137.

By the above operation, the image signal of the rectangular area shown in FIG. 27 is read out at high speed. Further, the charge accumulation time due to the above read operation is the time until the charge is transferred to the vertical shift register 133, and this time becomes the unit accumulation time t. This unit accumulation time ts is the burr measurement optical circuit 9.
The settings are made in the same manner as in the fourth embodiment based on the luminance data measured by No. 4. Then, the signal 'rIs song in the rectangular area is read out every single α accumulation time ts, and the BP
Manpower to F37. From now on, the explanation will be similar to the first embodiment,
Focus signal g(x), filtering signal f s (x)
is calculated and focus adjustment is performed.

As described above, according to this embodiment, the COD two-dimensional image sensor 1
By using 20, it is possible to adjust the focus to any desired area of the image at high speed and with high precision.

Note that if the image signals read from the CCD two-dimensional image sensor 120 are added column by column or row by column, the SN ratio can be improved and more accurate focusing adjustment can be achieved.

Next, as a sixth embodiment of the present invention, an automatic focusing device with an improved compensation calculation method will be described.

As explained in the second embodiment, the (n
) and (■) can be used to find the focal position. ΔX in equations (II) and (■) is the interval between three points, and is hereinafter referred to as an interpolation interval. Since this ΔX directly affects the focusing accuracy, it is necessary to select the optimum ΔX.

Furthermore, since the compensation calculation using the above three points also includes a margin error, it is necessary to correct this error.

Therefore, in this embodiment, the optimum interpolation interval is detected and the error in the interpolation calculation is corrected.

First, detection of the optimal interpolation interval will be explained.

The causes of errors in interpolation calculations include errors inherent in the calculation itself and errors due to noise in the image signal. The error of the calculation itself is a value determined by the interpolation interval ΔX and the detection positions of the three points (hereinafter referred to as interpolation positions). For example, if the focus signal is a Gaussian curve as shown in FIG. 28 (the closest point is x-0 and the infinity point is x-90), then depending on the interpolation interval and interpolation position, ? The percentage difference is as shown in FIG. In each figure, the vertical axis shows the error amount, and the horizontal axis shows the interpolated position. Note that L shown in each figure indicates the case where the two points on the right among the three points have the same error, as shown in FIG. 30(a). In addition, M in each figure is as shown in Figure 30 (b), when the errors on both sides of the three points are the same, R in each figure is as shown in Figure 30 (C), The two points on the left among the three points have the same error.

From FIG. 29, when Δx-14, the error due to the interpolation calculation itself becomes almost 0, and this value becomes the optimum value.

Further, errors due to noise in the image signal appear as noise in the focus signal, and a plurality of local maximum points occur in the focus signal. Therefore, in order to make the compensation calculation unaffected by noise,
It is necessary to make ΔX larger than a certain value.

By the way, as explained in the first embodiment, by filtering the focus signal, noise in the focus signal can be suppressed, and the condition for the minimum value of ΔX can be relaxed. That is, the larger the weight during filtering processing, the smaller the inter-capturing interval ΔX can be, and conversely, the smaller the weight, the larger ΔX needs to be. therefore,
The true optimal value of ΔX is determined by comparing the lowest value of ΔX determined from the weight of the filtering process and the noise of the image signal with the optimal value of ΔX that minimizes the error of the compensation calculation itself. Furthermore, since the error of the compensation calculation itself is influenced by the shape of the focus signal, it is determined by the MTF, focal length, F number, etc. of the BPF band 1 photographing optical system.

FIG. 31 shows a flow for optimal value detection under the above conditions. As shown in the figure, the noise of the focus signal is estimated from the weight of the filtering process and the brightness data of the image signal after passing through the BPF. Further, the shape of the focus signal is determined from the BPF band, focal length, F number, and MTF of the photographing optical system. Then, the optimal value of ΔX is determined from the estimated signal noise and the shape of the focus signal.

Next, correction of errors occurring in the sleeve opening calculation will be explained.

As shown in FIG. 29, a fixed error is added to the merging position calculated from the catching interval, ΔX, and the sleeve position. On the other hand, since the interpolation position can be calculated inversely from the values of the three points used in the interpolation calculation, the added error amount can be known. Therefore, by subtracting this error amount from the focus position determined by the interpolation calculation, the correct focus position can be obtained.

For example, the case of Δx-20 will be explained.

The amount of error due to the compensation calculation in the case of Δx-20 is as shown in FIG. 29, and this portion is extracted and shown in FIG. 32(a). Here, as shown in FIG. 32(b), the obtained focus signal is g(xi). g(x2), g(x3), is this the calculated focus position? Suppose that is X. In addition, Xt+40−Xi+20−X2, g (xi)
<g(x2), g(x2)>g(x3). g (xi). Each value of g(x3) is expressed as g(X
2), g + - g (xi)/g (x2). g3
Calculate g++g3 as 1g(x3) /g(x2)...(lO).Zo+g+gv...(1
l), then this 2. is the Z-axis value shown in FIG. 32(a), so it can be seen that the error is e(z.).

Then, the true focus position X is xT-X- e (zo) ・(1
2).

As described above, errors caused by the compensation calculation itself are corrected.

FIG. 33 shows the configuration of an automatic a-fish device to which the above-mentioned interrogation calculation method is applied.

Reference numeral 151 shown in the figure is an image sensor, and readout is performed by a drive circuit 152. In addition, ΔX
The values of the photographing optical system 3 are stored in a table and stored.
Focal length @f sent from 1, F number F1 Brightness data E sent from pre-metering circuit 94, BPF 37
:; area b1 Based on the filtering weight W of the filtering circuit 45, the optimum ΔX is searched and read out. The ROM 154 stores the error amount e(
Zr+) is stored in a table format, and ROM1
The corresponding error amount e(zo) is retrieved from the interpolation interval ΔX sent from the microprocessor 53 and the interpolation position Z sent from the microprocessor 150, and output to the microprocessor 150.

In the automatic focusing device configured as described above, when focus adjustment is started, preliminary photometry is first performed and the storage time of the image sensor 151 is set. Further, the brightness data of the subject measured by the blur photometry circuit 94 is output to the ROM 153. At the same time, the focal length @f of the photographing optical system 31, the F number F, and the band b of the BPF 37. Filtering weight W is R
Input to OM153. The ROM 153 calculates the optimum interpolation interval ΔX from these data by the microprocessor 15.
Output to 0.

Then, detection of the focus signal is started, and the filtering signal g (x) is incremented by 1 at intervals of X - ΔX. The values g(xi), g(X2) . . . of the three points near the maximum value of the filtering signal g(x) obtained in this way are: g
Using (x3), the microprocessor 150 executes capture and release calculations based on equations (II) and (■), and
x is calculated. Next, a correct focusing position XT is calculated based on equations (lO) and (11), and focusing is performed by moving the photographing optical system 31 to this focusing ridge position X7.

In this way, according to this embodiment, in the compensation calculation, the optimum interpolation interval ΔX is used and the error of the compensation calculation itself is corrected, so extremely accurate focusing adjustment can be performed. .

In addition, the optimum interpolation interval ΔX and the error of the interpolation calculation itself are tabulated and stored in the ROM 153 and ROMI 54, respectively.
Since the stored data is stored in the camera and the stored data is detected based on the imaging state, focusing can be adjusted extremely quickly.

In this embodiment, the band of the BPF 37 is fixed, but the same effect can be obtained by making the band switchable and making the interpolation interval ΔX constant.

〔Effect of the invention〕

According to the present invention, the focus signal is generated by adding the output amplitude values of the frequency components of the image signal, and this focus signal is further subjected to filtering processing, so that the S/N ratio of the focus signal can be improved and the focus signal is The maximum points of the eight signals can be suppressed. Therefore, focus adjustment can be performed at high speed and with high precision.

The image correction means is also provided, and the image signal read out from the image sensor is adjusted to a value corresponding to the accumulation nzzj interval h[i iF. This makes it possible to effectively prevent deterioration of the S/N ratio of the focus signal due to overexposure.

Since the readout time interval is set based on the brightness of one image formed on the image sensor, saturation of the image sensor can be reliably prevented and focus adjustment can be performed with high precision.

Furthermore, since an X-Y addressing type image sensor is used, focus adjustment can be performed on any area of the image.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the structure of the eye movement focusing device of the first embodiment, FIG. 2 is a diagram showing the structure of the line sensor, and FIG. 3 is a diagram showing the relationship between charge accumulation time and saturation in the image sensor. , Fig. 4(a) shows the rect function, Fig. 4(b) shows the spline function, Fig. 5(a) shows the spectrum of the rect function, and Fig. 5(b) shows the spectrum of the rect function. Figures 6(a) to 6(c) are diagrams showing the structure of the filtering circuit, Figure m7 is a diagram showing the focal signal and filtering signal, and Figures 8(a) and (b) are diagrams showing the spectrum of the spline function. A diagram for explaining the focus position detection direction, FIG. 9 is a diagram showing a focus signal and a filtering signal corresponding to the readout distance and focus signal detection time, and FIG. 10 corresponds to the frequency characteristics of a bandpass filter. Figure 11 (a) is a diagram for explaining the detection ability of
)(b) is a diagram showing the configuration of a modified part of the automatic focusing device shown in FIG. 1, FIG. 12 is a diagram showing the configuration of the automatic salt content device according to the second embodiment, and FIG. 13 is non-destructive readable. FIG. 14 to FIG. 16 are diagrams for explaining the difference calculation of focus signals, and FIG. 17 is a diagram for explaining the compensation calculation when detecting the in-focus position. FIG. 18 is a diagram showing a focus signal read out at a predetermined timing, FIG. 19 is a diagram of the third cusp embodiment, FIG. 20 is a timing waveform diagram for explaining the reset operation, and FIG. A diagram of the structure of the fourth length example,
FIG. 22 is a diagram showing a specific configuration of the area sensor, FIG. 23 is a diagram showing a rough area from the area sensor, FIG. 24 is a diagram showing the relationship between the readout clock and pixels, and FIG.
FIG. 26 is a diagram showing a partial modification of the device shown in FIG. 1, FIG. 26 is a diagram showing the configuration of the fifth embodiment, FIG. Figure 29 is a diagram showing the relationship between the amount of error and the interpolation position, and Figure 30 is a diagram showing the relationship between the amount of error and the interpolation position, and Figure 30 is a diagram showing the relationship between the amount of error and the interpolation position.
What is Figure 31, a diagram showing the values of three points at positions between i? +Ili
Figure 32 is a flowchart for detecting the optimum position using interval calculation, Figure 32 is a diagram for explaining the correction of error amount, Figure 33 is a configuration diagram of the sixth length example, and Figure 34 is a diagram showing the mountain climbing method. 35 is a diagram showing the relationship between the output amplitude of the image signal and the focus position, FIG. 36 is a diagram of the configuration of the automatic focusing device to which the phase correlation method is applied, and FIG. 37 is a diagram showing the configuration of the automatic focusing device applied. The figure shows local maximum points appearing in the focus signal. 31...Photographing optical system, 32...Line sensor, 34
, 72... Drive same path, 35... Counter, 37...
- Bypass filter, 38... Detector, 41... Divider, 42... Integrating circuit, 45... Filtering circuit, 46, 73... Microprocessor, 48...
・Motor drive circuit, 49...Pulse motor, 71...
・Non-destructively readable image sensor, 94...photometric circuit,
100, 120...area sensor. Atsushi Tsuboi, Patent Attorney, Atsushi Tsuboi Rec Ten Function 4th Figure Spectrum of the Rec 10 Function a a Spectrum of the Spline Function a ・Maf 冫 5 Figure 13 Figure 14 Figure 17 Figure 21 (a) Figure 31 Figure 32 (b) 2 Figure 37 Procedural amendments October 1, 1990

Claims (8)

[Claims] (1) An image sensor that captures an image formed by a photographic optical system, a drive means that changes the relative position of the image sensor and the photographic optical system in the optical axis direction, and a drive means that changes the relative position. means for capturing images at predetermined time intervals while reading out the charge accumulated in the image sensor as an image signal; and frequency extraction means for extracting frequency components in a predetermined band from the image signal read by the readout means. A focus signal generation means for detecting the output amplitude of the frequency component extracted by the frequency extraction means and adding the detected output amplitudes to generate a focus signal; A filtering means that performs filtering processing such as cutting off high frequency components included, and a calculation for detecting a focus state and a focus position using a plurality of values of a focus signal output from this filtering means, and control based on the results of this calculation. An automatic focusing device comprising: control means for outputting a signal to the driving means. (2) a non-destructively readable image sensor that captures an image formed by a photographic optical system; and a drive means that changes the relative position of the image sensor and the photographic optical system in the optical axis direction;
means for nondestructively reading out the charge accumulated in the image sensor as an image signal by capturing images at predetermined time intervals while changing the relative position by the driving means; Frequency extraction means for extracting frequency components of a band, and focus signal generation that detects the output amplitude of the frequency component extracted by the frequency extraction means and adds the values of the detected output amplitudes to generate a focus signal. and a focus signal outputted from the focus signal generating means, and performs focus state and focus position detection calculations using a plurality of values of the focus signal obtained by this difference calculation. An automatic focusing device comprising: control means for outputting a control signal based on a calculation result to the drive means. (3) When the charge accumulated in the image sensor reaches saturation, the charge accumulation is stopped, and the image signal read out at this time is corrected to a value corresponding to a predetermined accumulation time based on the time until saturation is reached. 3. The automatic focusing device according to claim 1, further comprising image signal correction means. (4) Two sets of points (Va, V
c) Set (Vb, Vd) on the focus signal, find a straight line m that intersects the line segment VaVc and the line segment VbVd and is parallel to the X axis, and find the point of contact between this straight line m and the line segment VaVc. is C1, and the point of contact between straight line m and line segment VbVd is C2,
Distance L2 in the X-axis direction between contact point C1 and point Va, and contact point C2
By setting the position of the straight line m in the Y-axis direction so that the distance L1 in the X-axis direction between points Vb and Vb are equal, and finding the midpoint M of the line segment C1C2 on the straight line m thus set, A focus position detection method characterized by detecting a focus position. (5) a non-destructively readable imaging element that captures an image formed by a photographing optical system; and a driving means that changes the relative position of the imaging element and the photographing optical system in the optical axis direction;
means for reading out charges accumulated in the image sensor as an image signal; and determining the brightness of the image from the image signal read by the reading means, and setting readout time intervals in the reading means according to the brightness. a photometric means; a frequency extraction means for extracting frequency components in a predetermined band from an image signal read out from the image sensor at readout time intervals set by the means; and an output amplitude of the frequency component extracted by the frequency extraction means. and a focus signal generating means for generating a focus signal by detecting and adding the detected amplitude values;
It is characterized by comprising a control means for performing a focus position detection calculation using a plurality of values of the focus signal generated by the focus signal generation means and outputting a control signal based on the calculation result to the driving means. Automatic focusing device. (6) When the brightness of the image is high as a result of the determination by the photometry means, the image sensor is reset every time an image signal is read out, and the control means controls the focus signal output from the focus signal generation means. When the brightness of the image is low as a result of the judgment by the photometry means,
The image sensor is reset when the peak value of the image signal to be output exceeds a threshold value set to a value smaller than the saturation value, and the control means is configured to control the difference between the focus signals output from the focus signal generation means. 6. The automatic focusing device according to claim 5, wherein a focus position detection calculation is performed using a focus signal obtained by the calculation. (7) X- to capture the image formed by the photographic optical system
a Y-address type image sensor; a drive means for changing the relative position of the image sensor and the photographing optical system in the optical axis direction; means for reading out electric charge at an arbitrary position accumulated in the image sensor as an image signal; photometry means for determining the brightness at the arbitrary position and setting a readout time interval of the image signal; and from the image signal read out by the reading means. Frequency extraction means for extracting frequency components in a predetermined band, and focus signal generation means for detecting output amplitudes of the frequency components extracted by the frequency extraction means and adding the detected output amplitudes to generate a focus signal. and filtering means that performs filtering processing such as cutting high frequency components included in the focus signal output from the focus signal generation means, and determining the focus state and the like using a plurality of values of the focus signal output from the filtering means. An automatic focusing device comprising: a control means for performing a focusing position detection calculation and outputting a control signal based on the calculation result to the driving means. (8) The automatic focusing device according to claim 7, wherein the X-Y addressing type image sensor is a CCD two-dimensional image sensor.