JPH03117276A - Automatic focus camera - Google Patents

Abstract

PURPOSE:To attain automatic focus with high focus accuracy even at low illuminance by inserting a high frequency cut-off filter cutting-off solid-state pattern noise from a solid-state image pickup element between the solid-state image pickup element and a focus evaluation value generating means. CONSTITUTION:An image pickup video signal outputted from an image pickup circuit 4 is amplified by a variable gain amplifier 21, inputted to a gain control circuit 22 so as to attain a reference level. When a gain control signal is larger than a prescribed value, a low illuminance state discrimination circuit 30 generates a low illuminance signal LOW as the low illuminance state and the signal is inputted to a focus evaluation generating circuit 35. The circuit 35 extracts only a luminance signal in the focus area by the operation of a synchronizing separator circuit and a gate control circuit and fixed pattern noise is eliminated from the luminance signal via an LPF. Thus, an offset component in the fixed pattern noise is not included during the focus evaluation and focusing is applied by the focus evaluation depending only on the picture contrast even at low illuminance.

Description

DETAILED DESCRIPTION OF THE INVENTION (A) Field of Industrial Application The present invention relates to an auto-7 focus device for a camera that automatically aligns the focus based on a luminance signal in a captured video signal obtained from an image sensor.

(b) Conventional technology The method of using the high-frequency components of the video signal itself from the image sensor to evaluate focus control in the auto-7 orcus device of a camera is essentially free of variation and has a limited depth of field. It has many advantages, such as being able to focus accurately even when photographing shallow or distant objects. Furthermore, there is no need for a special sensor for autofocus, making it extremely simple.

Conventionally, an example of this autofocus method is “NHK
It is described in "Technical Report" S40, Vol. 17, No. 1, Volume 86, Page 21, as "Automatic focus adjustment of video camera using hill-climbing servo system" by Ishida et al., and is known as so-called climbing servo control. This mountain-climbing servo control is a method in which the lens is moved back and forth in the optical axis direction so that the amount of high-frequency components of the video signal is always maximized.

Japanese Patent Application Laid-Open No. 63-125910 (GO2B7/11)
discloses an example of the above-mentioned so-called mountain climbing autofocus system. Here, regarding this prior art, the second
The outline will be explained using the diagram and FIG. 3. FIG. 2 is an overall circuit block diagram of the conventional technology. In this figure, an image formed by a lens (1) is turned into a video signal by an image pickup circuit (4) including an image pickup element, and a focus evaluation value generation circuit ( 5). Focus evaluation value generation circuit (5
) is constructed as shown in FIG. The vertical synchronization signal (V
D), the horizontal synchronization signal (HD) is input to the gate control circuit (5b) to set the sampling area as 7 orcus areas. The gate control circuit (5b) sets a rectangular sampling area in the center of the screen based on the vertical synchronization signal (VD), the horizontal synchronization signal (HD), and the fixed oscillator output that drives the image sensor. A gate opening/closing signal that allows passage of luminance signals only within the range is supplied to the gate circuit (5C).

Only the luminance signals corresponding to the seven orcus areas are filtered by the gate circuit (5c) through the high-pass filter (l(
, P, F) (5d), only the high frequency components are separated, and the amplitude is detected in the next stage detection circuit (5e). This detection output is converted into a digital value by the A/D conversion circuit (5f) at a predetermined sampling period, and is sequentially sent to the integrator (5g).
is input.

Specifically, this integrator (5g) consists of an adder that adds A/D conversion data and latch data of a subsequent latch circuit, and a latch circuit that latches this added value and is reset every field. It is a so-called digital integrator consisting of
The sum of all A/D conversion data for one field period is output as a focus evaluation value. Therefore, the focus evaluation value generation circuit extracts the luminance signals within the seven orcus areas in a time-division manner, digitally integrates this high frequency component over one field period, and uses this integrated value as the focus evaluation value for the current field. It will be output. Immediately after the auto 7 orcus operation starts, the first focus evaluation value is stored in the maximum value memory (6)
and is held in the initial value memory (7). Thereafter, the focus morph control circuit (10) moves the lens (1) forward and backward in the optical axis direction.
3) in a predetermined direction and the second comparator (9)
Monitor output. The second comparator (9) compares the focus evaluation value after driving the focus motor with the initial evaluation value held in the initial value memory (7) and outputs the magnitude thereof.

The focus motor control circuit (10) includes a second comparator (9).
) in the first direction until it produces an output of greater or lesser.
If the Orcasmoke (3) is rotated and an output is made that the current focus evaluation value is larger than the initial evaluation value and larger than the preset fluctuation range, the direction of rotation is maintained as it is, If it is output that the current evaluation value is smaller than the above fluctuation range compared to the initial evaluation value, the direction of rotation of the 7 Orcasmoke (3) is reversed, and the rotation direction of the first comparator (8) is Monitor output.

The first comparator (8) compares the maximum focus evaluation value held in the maximum value memory (6) with the current focus evaluation value, and the current focus evaluation value is determined by the maximum focus evaluation value stored in the maximum value memory (6). There are two types of comparison signals (1st mode) that are larger than the content (1st mode) and 2nd mode that are reduced to more than the preset first darkness value (2nd mode).
Pi) (P2) is output. Here maximum value memory (6)
is updated based on the output of the first comparator (8) if the current focus evaluation value is larger than the content of the maximum value memory (6), and is always updated to the maximum value of the focus evaluation value up to now. The value is retained.

(13) is a focus ring (
2) This is a motor position memory that stores the 7-occasion ring position upon receiving the 7-occasion ring position signal that indicates the position of 1st comparison 5 (8), similar to the maximum value memory (6).
Based on the output of , the focus ring position is updated so as to always maintain the focus ring position when the maximum evaluation value is reached. Here, it is a well-known technique that the 7 orcus ring (2) is rotated by the 7 orcus smoke (3), and the lens (1) moves forward and backward in the optical axis direction in accordance with this rotation.

Note that the 7th place ring position signal is output by a potentiometer that detects the 7th place ring position, but the 7th place smoke (3) is used as a stepping motor, and the amount of rotation of this motor toward the near point or It is also possible to use positive and negative step amounts and express the position of the focus ring or focus morph using these step amounts.

The focus motor control circuit (10) includes a second comparator (9).
) focus morph in the direction determined based on the output (
3), the output of the first comparator (8) is monitored, and in order to prevent malfunctions due to noise in the evaluation value, the current evaluation value is set to the maximum evaluation value at the output of the first comparator (8). In comparison, the second mode in which the width is smaller than the preset first threshold value (Δy) is instructed (reaches Q in FIG. 4), and at the same time the focus morph (3) is reversed. After this reversal, the contents of the position memory (13) and the current focus ring position signal are compared in the third comparator (14), and when they match, that is, the focus ring (2) has the maximum focus evaluation value. The 7-o-ka-smoke control circuit (10) functions to stop the 7-o-ka-smoke (3) when it returns to the position (P). At the same time, the focus motor control circuit (10) outputs a lens stop signal (LS) to complete the focusing operation. Incidentally, a flowchart when the above-mentioned series of focusing operations are processed by software using a microcomputer is shown in FIG. 5(a).

When the subject moves after this focusing operation is completed, the focus evaluation value changes as shown by the dashed-dotted line in Fig. 6, and the lens position where the value reaches its peak also changes, and the focusing operation is resumed and the 7 orcus memory motor is activated. You will need to reboot. Therefore, when the lens stop signal (LS) is generated, the fourth memory (
11) stores the focus evaluation value at this point in time, and compares this value with subsequent focus evaluation values in the fourth comparator (12), and calculates the difference between the two as a preset second threshold ( M), the focusing operation is restarted again even though the subject has moved. Incidentally, a flowchart regarding this restart is as shown in FIG. 5(b).

In the prior art, if the second threshold value (M), which is a guideline for restarting the focusing operation, is set to a fixed value, there will be a large difference in the restart timing depending on the high frequency component of the subject. In other words, the peak of focus evaluation values for a high-contrast subject with a vertical black and white stripe pattern that contains many high frequency components is a steep peak with a high peak (Fig. 6(b)). In the case of a low-contrast object such as a white wall where high-frequency components are difficult to produce, a gentle mountain with a low peak is formed (FIG. 6(a)). Therefore, if the second threshold (M) is a fixed value, a slight displacement of the subject will cause a large change in the focus evaluation value in the former case, and the focusing operation will be restarted immediately; however, in the latter case, the focusing operation will be restarted immediately. Even if the subject moves considerably, the focus evaluation value will not change significantly and the focusing operation will not be restarted unless the subject becomes considerably out of focus.

Therefore, the above-mentioned problem can be solved by setting the second threshold, which serves as a guideline for determining when to restart, to a value proportional to the focus evaluation value (Fo) at the time when the lens stop signal (LS) is issued. It can be solved. For example, M = F o X 0, 0 5 + f e
... By setting (1), the sixth
In the case of figure (b), the second threshold value (M) is large, and in the case of (a), the second threshold value can be made small, and in either case, the focusing operation is restarted when the subject is displaced by a certain amount. It turns out.

(C) Problems to be Solved by the Invention Video cameras generally have various types of noise, but solid pattern noise (FPN) is a noise specific to video cameras using solid-state image sensors. Fixed pattern noise occurs as a result of non-uniformity due to various factors during the manufacturing process of solid-state imaging devices, and appears as pixel defects, uneven sensitivity, uneven dark current, and the like.

Solid-state imaging devices with pixel defects are considered defective and are not used in video cameras, but fixed pattern noise due to uneven sensitivity cannot be avoided in video cameras using solid-state imaging devices.

With a video camera, fixed pattern noise is almost undetectable during general shooting, but when the gain of the signal system increases, such as during low illumination, the fixed pattern noise increases in proportion to the gain. , it becomes possible to detect it as noise.

Fixed pattern noise is sensitivity unevenness for each pixel of a solid-state image sensor, and the amount of high frequency components due to this noise is constant for each field. Therefore, during normal shooting, Figure 6 (a)
) (b), the focus evaluation value changes, but at low illuminance, the fixed pattern noise component increases due to the increase in the gain of the video signal, and the focus evaluation value changes as shown in Figure 7 (a) and (b). The value will start changing. Therefore, when an object that originally has a low peak and a gentle mountain is photographed under low illumination, the focus evaluation value is shifted upward, and although the focus evaluation value itself is high, the slope remains gentle. Therefore, the formula (1
), when the second threshold value is set in (a) of FIG.
As in b), the second threshold (M) is set large. However, since the slope of the mountain itself does not change, even if the same subject is photographed at the same distance, it will take time to confirm the subject's movement in low light due to the influence of FPH, and the subject will move significantly If the image does not reach an out-of-focus state, it will not be restarted.

Note that the shaded portions in FIGS. 7(a) and 7(b) are offsets for fixed noise components due to fixed pattern noise. Also, although it is omitted in the circuit block diagram of FIG. 2, an AGC circuit is usually inserted after the imaging circuit (4), and when the imaged video signal level is low, such as during low illuminance, It functions to increase the gain and increase the level.

(d) Means for solving the problem Fixed pattern noise is due to sensitivity unevenness for each pixel of a solid-state image sensor, so the frequency components of the noise are the horizontal transfer frequency for driving the solid-state image sensor and the color signal. Depends on the filter array to get . Therefore, when the illuminance is low, an LPF that blocks the noise frequency component caused by the fixed pattern noise is inserted before or after the HPF for extracting the high frequency component of the luminance signal.

(E) Function Since the present invention is constructed as described above, the focus evaluation value obtained at low illuminance does not include a fixed pattern noise component, and the influence of this fixed pattern noise on the focusing operation is eliminated.

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

In each figure, the same parts as in FIGS. 2 and 3 are denoted by the same reference numerals, and explanations thereof will be omitted.

FIG. 1 is an overall circuit block diagram of this embodiment. In the figure, (20) is an AGC circuit composed of a variable gain amplifier (21) and a gain control circuit (22), and an imaging circuit (
4) The captured video signal output from this AGC circuit (
20), the abnormally high level or abnormally low level can be suppressed. Specifically, a variable gain amplifier (2
The imaged video signal amplified by the gain (G) in step 1) is integrated by an integrating circuit (not shown) with a sufficiently large time constant, and then input to the gain control circuit (22) to obtain a predetermined signal. The signal is supplied to a variable gain amplifier (21) that controls the gain (G) so that it reaches the reference level.

(30) is a low-light state discrimination circuit that determines that the subject is in a low-light state when the gain control signal is larger than a predetermined value (V) and generates an H-level low-light signal (t, OW >.Here, The predetermined value (V) is a value that serves as a guideline for a low illuminance state, and in a low illuminance state where the AGC gain must be increased to exceed this predetermined value (V), the level of FPN becomes significantly large.

The imaged video signal and low illumination signal (tOW>) that have been appropriately amplified by the AGC circuit (20) are processed by the focus evaluation value generation circuit (
35).

The focus evaluation value generation circuit (35) is constructed as shown in FIG. Synchronous separation circuit (5a) and gate control circuit (5
By the operation b), only the luminance signals corresponding to the seven orcus areas at the center of the screen are extracted from the gate circuit (5C), as in the conventional example. This gate output is supplied to the fixed contact (36a) of the switch circuit (36) and
It is supplied to the fixed contact (36b) via P F (37).

Switching control of the switch circuit (36) is performed using a low illuminance signal (LO
W), and when the low illuminance signal (LOW) is at L level, that is, at normal gain, it switches to the fixed contact (36a) side, and the HP F (5d), detection circuit (5e), The integrator (5g) passes through the A/D converter (5f).
) outputs a focus evaluation value, and based on this, a hill-climbing focusing operation is performed in a subsequent circuit.

On the other hand, if the subject illuminance is extremely low, the AGC circuit (
20) increases significantly and exceeds a predetermined value (V), the low illuminance signal (LOW) becomes H level, the switch circuit (36) switches to the fixed contact (36b) side, and the brightness within the focus area increases. The signal is input to L P F (37).

FIG. 9 shows this L P F (37) and HP F
The passage rate characteristic of (5d) is shown, and HP F (
5d) has a cutoff frequency of, for example, 100 Klb or more in order to extract high-frequency components that occur in the luminance signal as it approaches the in-focus state. L P F (37
) has a cutoff frequency below the frequency of the noise spectrum due to fixed pattern noise, so this L
Fixed pattern noise is removed from the luminance signal that has passed through P F (37).

In this way, a brightness signal from which the fixed pattern noise has been removed is generated at the fixed contact (36b), and when this passes through the HP F (5d), a brightness signal in the band shown in FIG. 9(A) is finally extracted. Therefore, the focus evaluation value output from the integrator (5g) does not include the offset of fixed pattern noise as shown by the diagonal lines in Figure 7, and even in low illuminance, the screen contrast remains unchanged as shown in Figure 6. A focusing operation is performed using a focus evaluation value that depends only on the .

Note that the fourth comparator (40) for checking the movement of the subject uses the focus evaluation value at the time of generation of the lens stop signal (LS) stored in the fourth memory (11), as described in the conventional example. It functions to monitor whether the difference between the current focus evaluation values exceeds the second threshold (M). Here, the second threshold (M) is calculated by the formula (
1), it changes in proportion to the size of the content held in the fourth memory (11).

By the way, in a solid-state image sensor used in a single-chip color video camera, a color separation filter is pasted on each pixel of the image sensor in order to obtain color information from one solid-state image sensor. This color separation filter array includes a stripe array,
There is a mosaic arrangement; for example, a stripe filter using primary color filters is shown in FIG. 10(a), and a mosaic filter is shown in FIG. 10(b). Signal processing of a solid-state image sensor with stripe filters bonded will be described below.

As shown in Figure 11, in a solid-state image sensor with RGB stripe filters pasted together, charges photoelectrically converted by multiple photosensors (+)h) arranged two-dimensionally are vertically shifted for each field. transferred to the register (VR),
Further, the data is vertically transferred column by column within this vertical shift register and transferred to a horizontal shift register (HR). Furthermore, in synchronization with the horizontal transfer pulse (IP), the horizontal shift register (
The charges that are sequentially transferred from l(R) to the output section (OUT) and then transferred from the vertical shift register to the horizontal shift register are sequentially output as imaging signals over one horizontal scanning period.

This imaging signal is processed by a CD5 (correlated double sampling) circuit in the imaging circuit (4) to remove CCD-specific noise and convert it into an R, G, and B point sequential signal as shown in FIG.
, B using three sample and hold circuits for R, G, and B.
The individual color signals of are separated. This sample hold period is the interval at which the same color components can be taken out, and in the case of this stripe filter, R, G, and B are lined up sequentially, so
This corresponds to three periods of horizontal transfer pulses. The video signal is then created by processing the sample-and-hold output of each color component.

Therefore, even if fixed pattern noise occurs due to sensitivity unevenness that exists in any of the photosensors, the frequency (f,) of the fixed pattern noise will be fv = fo X 1/N (N: integer)...(2)
becomes. Here, fo is the frequency of the horizontal transfer pulse (horizontal transfer frequency), the ratio of the period in which the same color component of the dot sequential signal occurs to the period of the horizontal transfer pulse (N = 3 for the stripe filter), and f for the stripe filter. =f
, /3.

Also, in a mosaic filter, G and B or G and R
are lined up repeatedly, so N=2, and f,=f,
/2.

As mentioned above, the frequency of fixed pattern noise changes depending on the filter arrangement as shown in equation (2), so the cutoff frequency of L P F (37) is changed according to the arrangement of color filters used, It is necessary to ensure that fixed pattern noise always falls within the cutoff frequency band.

A flowchart for processing each circuit operation of the above-described embodiment in software using a microcomputer is shown in FIG.

In this example, during the actual focusing operation, the lens (1) is moved forward and backward in the optical axis direction using 7 orcasmoke (3) to change the distance between the lens (1) and the solid-state image sensor. However, it is also possible to fix the lens (1) and move the solid-state image sensor itself forward and backward in the optical axis direction using a motor or bimorph controlled by the output of the seven-channel smoke control circuit (10). In any case, there is no problem as long as the relative position of the lens with respect to the solid-state image sensor is changed.

(G) Effects of the Invention As described above, according to the present invention, it is possible to suppress the fixed pattern noise from being mixed into the focus evaluation value that is the basis of focusing operation, and to achieve autofocus with high focusing accuracy even in low illuminance. can be realized.

[Brief explanation of drawings]

1, 8 to 13 relate to an embodiment of the present invention, FIGS. 1 and 8 are circuit block diagrams, and FIG. 9 is an LPF,
Transmittance characteristics of HPF, Figure 10 is an explanatory diagram of the color separation filter arrangement, Figure 11 is an explanatory diagram of CCD drive, and Figure 12 is CCD.
The timing chart of OD output, and FIG. 13 is a flowchart. Also, Figures 2 and 3 are flowcharts of the conventional example, Figure 4 is an explanatory diagram of the operation of mountain climbing autofocus, Figure 5 is a flowchart showing the circuit operation of Figure 2, and Figures 6 and 7 are threshold values. FIG. 3 is an explanatory diagram of settings. (5d)...HPF, (35)...Focus evaluation value generation circuit, (10)...7 focus motor control circuit, (3
7)...LPF0

Claims (1)

[Claims] (1) a low-pass cut filter that extracts high-frequency components of a luminance signal obtained from a solid-state image sensor; and a focus evaluation value generating means that integrates the output of the low-pass cut filter for a predetermined period and outputs it as a focus evaluation value; An autofocus camera comprising a focus control means for performing focus control so that the focus evaluation value is maximized, the solid-state pattern noise from the solid-state image sensor being interposed between the solid-state image sensor and the focus evaluation value generation means. An autofocus camera characterized by inserting a high-frequency cut filter. (2) The autofocus camera according to claim 1, wherein the high-pass cut filter cuts off components having a frequency equal to or higher than 1/N (N is an integer) of the horizontal transfer frequency of the solid-state image sensor.