JPS6383621A - Deviation detector - Google Patents

Abstract

PURPOSE:To obtain a detector capable of warning or removing a wrong deviation measurement by deciding whether or not a calculated phase difference accurately indicates a deviation quantity. CONSTITUTION:The signal of the sensor device 12 consisting of sensor arrays 12a and 12b and a CCD is inputted to a processor 11 through an A/D converter 13. Further, a deviation quantity display device 17 and an error display device 12 operate with the control output of the device 11. The sensor device 12 is operated by a CCD driving device 15 and further a clock pulse signal is sent from a clock generator 16 to the devices 15 and 16. Then the correlation between the output signal pattern A of the device 12 which photodetects part of luminous flux passing a photographic lens at a certain point of time and the corresponding output signal pattern which is obtained a certain time later is checked, and sampling operation for each pixel is performed based on the correlation quantity; and the correlation quantity of the sampled signal is found to calculate the deviation quantity, which is compared with that which is found by the last correlation, thereby judging and error in the detected deviation quantity.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a camera shake detection device that detects shake using a focus detection optical system and a processing system.

(Prior art) The principle of detecting camera shake using a focus detection device based on a phase difference installed in a camera is, for example, disclosed in Japanese Patent Application Laid-open No. Shofi.
It is known from Japanese Patent No. g-166910.

1 to 3 show the principle of focus detection using a known phase difference that can be utilized in such a known blur detection method. 1st
The figure shows the in-focus state, FIG. 2 shows the front focus state, and FIG. 3 shows the rear focus state. As shown in FIGS. 1 to 3 (A), an optical image formed by a light beam passing through two different areas of the objective lens 1 is transmitted to two sensor arrays 3a, 3b and output signal Sa.
.

sb is obtained, and the focus state of the photographic lens can be determined from the phase difference between Sa and Sb. The blur detection method known from the above publication detects one of the output signals Sa (or sb) twice with a fixed time difference, and detects blur from the phase difference between the two.

At this time, the following algorithm is known for determining the phase difference d. Sensor 3a at a certain point in time
The signal Sa from the sensor 3a and the signal Sa' from the same sensor 3a after a certain period of time are respectively converted into A image (a(+1 + af21
, "" a+n+ ), A' image (a' Nl , a
′+2+ , ” ' +Nl ) N ◆
Let the a image signal be the correlation amount PK of the A image and A' image (K<0) (K2O) (where n=N-IKI.

−N/2≦ and ≦N/2). According to (1), (+)', the correlation ffi
PK calculates the consistency between the A image and the A' image while changing the phase, and the K that minimizes the correlation fit PK becomes the above-mentioned phase difference d. Therefore, by performing the calculation of equation (+) and (1)', information regarding the amount of movement of the image due to shake can be obtained. FIG. 4 shows examples of images A and A'. If there is no blurring, the output signals will be almost the same as shown in FIG. 4(A), and the PK will be as shown in FIG. 5 α, so the phase difference d will be 0. When blur occurs, the subject pattern imaged on the sensor changes, so the image shown in Figure 4 (B),
As shown in (C), the waveform becomes substantially round. The waveform correlation amount PK at this time takes the form as shown in FIG. 5 β and γ, and the phase difference d can be determined from the minimum value of PK.

As another algorithm for determining the phase difference d, A image and A
A method of sampling a signal above a certain threshold level is also known (Japanese Patent Application Laid-Open No. 17492-1982).
(See Publication No. 8).

(Problems to be Solved by the Invention) However, these known methods cannot be used in cases where the phase difference is so blurred that it is essentially impossible to detect it, or when there is a scene and corresponding blur that makes it easy to miscalculate the phase difference. This included the possibility that the phase difference d would be calculated as an erroneous amount of shake different from the original amount of shake.

An object of the present invention is to provide a blur detection device that can determine whether the calculated phase difference d accurately represents the amount of blur and can warn or eliminate erroneous blur measurements.

(Means for Solving the Problem) In a camera shake detection device that detects shake using an optical system for focus detection and a processing system, there is a sensor device (12) that receives a part of the light beam of a photographing lens. The correlation between the output signal pattern A at the time and the output signal pattern A' after a certain period of time is examined, each pixel is sampled based on the correlation amount, the correlation amount is determined for the sampled signal, the blur amount is determined, and the The object of the present invention is to have means for determining whether the detected amount of blur is incorrect by comparing it with the amount of blur determined by correlation.

(Embodiment) The embodiment of the present invention shown in FIG. 6 will be described in detail.

FIG. 6 shows a camera control device that can also be used for focus detection. Engineering 1 is the CPU (central processing unit).

The image signal processing device is composed of, for example, a one-chip microprocessor having a memory circuit, one output port, and the like. This processing device 11 includes a sensor array 12a.

12b and a CCD (charge-coupled device), a signal from the sensor device 12 is input manually via an A/D converter 13.

In addition, the control output of the control device 11 causes the blur amount display device 17 to
and the error display device 21 is activated. The sensor device 12 is operated by a CCD drive device 15, and a clock pulse signal is sent to the CCD drive device 15 and the processing device 11 from a clock generator 16.

Two images are formed on the sensor arrays 12a and 12b by the light beams that have passed through different areas of the photographic lens, and a control signal Φ from the CCD drive device 15 is generated. .

Accumulation and transfer of optical images are performed by SH and ICG. That is, when the processing device 11 supplies the start signal ST to the drive device 15, the drive device 15 receives the signal CLK from the clock generator 16.
The clock Φ generated by

At the same time, an accumulation start signal ICG is sent to the sensor device 12. The sensor device 12 starts accumulating two images from this point, and sends an accumulation completion signal EOI to the drive device 15 when a predetermined accumulation level is reached. The drive device 15 transmits the charge conversion output transfer signal SH to the sensor device! 2, the charge accumulated in the sensor device 12 is transferred from the sensor section to the CCD section, and at the same time, an end signal END is sent to the processing device 11. Subsequently, the clock Φ from the drive device η15. In synchronization with
The A/D converter 13 outputs the 8-bit A/D converter 13 in synchronization with the conversion command signal ADC from the drive device I5.
/D conversion is performed and the digital time series signals DO to D7 are sent to the processing device 11. The processing device 11 generates a manually generated two-image signal A (a(ll + af21 !”” a
fnl ) 1B(b(11+b(21+...b
) is stored in memory temporarily. When performing known automatic focus detection, the phase difference between the two images A and B can be used as a measure of the in-focus state. When detecting camera shake, a series of supplementary sequences starting from sensor image accumulation are repeated at regular time intervals as described above. That is, the signal A' of the previous sequence and the signal A of the next sequence
or temporarily stored in memory.

If there is no blur, the signals A' and A will be exactly the same, so the phase difference d between the two signals will be zero. On the other hand, when there is a shake, the phase difference d does not become zero, and it is possible to use this as a measure of the amount of shake. The phase difference d is shown below [I], [II], [[
II].

[1: ] A +++ = a (+1B (il
"A ' N) (i=1.2....N) (K2O) However, n=N-IKI.

(-N/2≦5N/2) [[[] d=K l m1n(P,,)(-N
/2≦≦N/2) Formula [IIr] means “d is K at which PK becomes minimum”.

However, the phase difference d thus calculated may not indicate the correct amount of blur depending on the photographed scene or the degree of blur. Therefore, it is necessary to determine whether the calculated movement difference d is correct as a blurring river. The procedure for such determination is shown below.

First, the phase difference d between A and B is determined by the method described above. Then, the difference X between the corresponding elements of A and H in the phase difference d is determined by the following equation.

[rv] Xi = + Bfi or 1dll
-A+i+ l(d<O) X + = l B ++, A +++d+
l (d≧0)i=1. 2. ... n.

n=N-1dl Furthermore, Xi <p□. When a combination of elements A and H that satisfies /n occurs, 1 is written in the tag memory.

That is, [V] tag memory of element of A MarkA,,)
= 1j=i (d<0) j=i+a (d≧0) Tag memory of element of B MarkB (11= 11=
(i+Idl) (d<0) 1=i (d≧0) (i=1.2...n) When a combination of elements A and H that satisfies X1≧P□dt/n occurs, tag Write 0 to memory. That is, [■ Tag memory of element of A MarkA (J) = 0j=i
(d<0) j = i + d (d≧0) Mark 8 tag memory of element of B, =01=(i
+ l d l ) (d<0)1=i
(d≧0) Next, calculate PK for each of the four possible states as shown below.

■MarkA (J, = A set as O (jl +
Addition is performed only when the combination B+m) is obtained. That is, calculation is performed using the following formula.

PK□=± l B +-+ A tJ, lHowever, j
=i, m=i+lKl (K<0)j=(i+K),
m=i (K2O)n=N −l K l , −
N/2≦≦N/2 From this, the phase difference d, is determined by d, =Kl m1n(PK,) (-N/2≦≦N/2).

(2) Similarly, addition is performed only when the combination of MarkB + 1111 = 0 (Bfml + A fJ) is obtained, PK2 is obtained, and d2 is calculated.

■Similarly, MarkA (, A with H=1 set,,
,,.

Addition is performed only when the combination of B+ml is obtained, PK3 is obtained, and d3 is calculated.

■Similarly, MarkB (AL set as ml = 1)
J).

Addition is performed only when the combination B(1) is obtained, PK4 is obtained, and d4 is calculated.

Then, the maximum value d n+ax and the minimum value d of d1 to d4
□. If the difference d max dmin≧e holds true (e is a positive value), it is determined that the calculated phase difference d does not accurately represent the amount of blur.

In order to confirm d calculated as above, sampling is performed for the large A element, the small A element, the large B element, and the small B element with the difference x in the phase difference d, and the phase difference for each is When calculating d, -d4, if the calculated d is inaccurate or is likely to contain an error, d, ~d, will have different weight values, so (
The accuracy of d can be determined by the value of d, , 1aX-d, .

The processing device 11 calculates the phase difference d corresponding to the amount of blur from Q4 to Q.
1 terminal outputs H or L digital code information and lights up the LED 17 according to the code value. Also, if d is determined to be incorrect, the terminal ERR
It outputs H and lights up the LED21.

The operation of the apparatus of the present invention having the above configuration will be explained with reference to the flowchart shown in FIG. The processing device 11 is preprogrammed with an algorithm based on the flowchart shown in FIG. Note that the numbers in brackets [ ] represent step numbers.

[1] The processing device 11 issues a start signal ST, and the CCD drive device 15 sends a signal ICG to the sensor device 12 to start accumulating an optical image.

[After two predetermined accumulation levels are reached, the sensor device 12
The drive device 15, which has received the accumulation completion signal EOI from
The optical image conversion signal O8 of the optical image projected above is A/D converted, and the processing device 11 receives the digital signal a(11+b+11(i=1...) of the two images A and B.
2. ...N) manually and sequentially store them in memory.

[3] A14 to Bz+. Assign the values of A, 1. ni a
Assign , , . (The above [■]) (For the first time, calculate B (the value of il is indeterminate.) [4] d. (The above [II], [!II]) [5] MarkA +11, MarkB L+
Generate +.

(Said [IV], [V], [VI]) [6]
d+, d2. Calculate d:+, d4.

(Above, ■, ■, ■) [7] d+, ct2. The minimum value of ct3.64 is d mio, and the maximum value is d. Let it be ax.

[8] In the processing device 11, d ff1ax d
The value of +min and the error tolerance value e+ (positive value) are compared.

[9] If dffiax d 1m+n < e +, it is determined that the error is allowed, the ERR terminal is output, the LED 21 is turned off, and the error display is turned off.

[10 d max dmi. <e, if not,
As the error tolerance is exceeded, output high to the ERR terminal,
Light up LED 21 and turn on the error display.

[11] The value of d is output as a binary code to Q+ to Q4, and the LED 17 is turned on or off.

As mentioned above, the values of d, ~d4 and d do not show the correct values corresponding to the blur in the first calculation after the power is turned on, and 2
The calculated value from the first time is valid.

In this embodiment, the selection of pixels was determined by the magnitude relationship of Xl with respect to the average value PK(./n) of xl at the phase difference d, but it is also possible to select some pixels in order of increasing or decreasing Xl. In this case, there is no dependence of the number of pixels selected on the scene, so stable error detection is possible.

As explained above, in the phase difference d calculated for the entire pixel, pixels can be distinguished based on the magnitude of the difference between the values of corresponding pixels, the phase difference for each pixel can be calculated independently, and the results can be compared. Therefore, it is possible to determine whether the above-mentioned phase difference is accurate or not, which has the effect of preventing erroneous detection of image blur.

FIG. 8 shows PK,d calculated using the conventional method and PKl, d+, PK2°d2. PK3
．． d*, PK4. This is a diagram illustrating d<. As can be seen from Fig. 8, the phase difference d can be calculated even for scenes where the phase difference cannot originally be calculated using the conventional method.
It was difficult to determine whether this phase difference d correctly indicated image blur. Using the method according to this invention, PK
When calculating I * P K 2 + P K3 + P K4, d, , d2 and d3 . as shown in FIG. d
4 is different, and it can be seen from the value of d maN-d nain that the obtained d does not represent the correct shake.

Figure 9 shows that the 8th
This figure shows the results of calculations similar to those shown in the figure. In this case, the d rash amount is displayed correctly, so d
, d+ , d2. d3. It can be seen that d4 matches.

As described above, for example, d□ax dmi. By comparing the value of , with a predetermined value, it becomes possible to issue a warning of incorrect calculation of the amount of blur, etc., which is effective for image measurement, image stabilization by feedback of blur information, and the like.

[Brief explanation of the drawing]

Figures 1 to 3 are principle diagrams of automatic focus detection, Figures 4 and 5.
The figure is an explanatory diagram of shake detection, Figure 6 is a block diagram of the processing system of the camera shake detection device according to the present invention, Figure 7 is a flowchart of the processing system, and Figures 8 and 9 are values of PK to d. FIG. 11...processing device, 12...sensor device, 13
... A/D converter, 15... Drive device, 16.
... Clock generator, 17... Shake π display device,
21...Error display device patent applicant Canon Co., Ltd. Agent Yukihiro Takanashi
1 Figure (A) (1 Akira 2 times (A
) (9) Figure 3 (A)
(9) Helmet ・4 [koCA] (B) (C) 5th Akira 7 Roko 8 [2] (A) 11111-1---k (B) - to □ d3 (lj4 Akira 9 2 (c・) K (e) ~ □

Claims (4)

[Claims] (1) In a camera shake detection device that detects shake using an optical system for focus detection and a processing system, the output signal pattern A at a certain point in time of the sensor device (12) that receives a part of the light flux of the photographic lens. A blur detection device comprising means for sampling an output signal pattern A' after a predetermined period of time has elapsed, determining a correlation amount for the sampled signal, and determining an amount of blur. (2) A means for sampling using at least two sampling methods and comparing the amount of blur calculated for each method, and determining that the calculated amount of blur is incorrect if the two are significantly different. The device according to claim 1, characterized in that it has: (3) The sampling means successively compares the signals of A and A' with a phase difference that has a good correlation with the entire pattern of A and A', and the absolute value of the difference between A and A' among the signals of A and A'. By comparing A with the average of the absolute values of the differences, the following processing is performed: (a) For the part where the absolute value of the differences is larger than the average value, A
(b) For the part where the absolute value of the difference is larger than the average value,
(c) For the part where the absolute value of the difference is smaller than the average value, sample the pattern of A
(d) For the part where the absolute value of the difference is smaller than the average value,
2. The apparatus according to claim 1, wherein the apparatus performs at least one of sampling a pattern of . (4) forming a pattern related to an object image whose spacing changes depending on the focusing state of the objective lens; receiving each of the patterns with a sensor array having a plurality of pixels; A phase difference is determined by determining the correlation based on the signal and the signal related to the same pattern after a certain period of time has elapsed, and pixels are distinguished based on the magnitude of the difference between the values of the corresponding pixels, and each pixel is A blur detection system comprising the step of independently calculating a phase difference and comparing the results to determine whether or not the phase difference is appropriate, and detecting blur based on the magnitude of the phase difference. Method.