JPH05107619A - Camera capable of detecting/correcting blurring - Google Patents

Abstract

PURPOSE:To provide a camera which can accurately detect/correct blurring even on a photographing lens whose focal distance is great, and further, detect/ correct the blurring even on an object having low brightness, as well. CONSTITUTION:A photographing lens circuit 7 outputs the focal distance information of a lens fitted thereto, to a main CPU 1. The main CPU 1 determines the periods of the detection/correction of the blurring, based on the focal distance information outputted from the photographing lens circuit 7. A blurring detecting/correcting means 12 executes calculation to obtain blurring data from the image pickup data of a CD, and corrects the blurring by controlling a correcting optical system through the use of a control CPU. After that, the detection/correction of the blurring are repeated in the periods determined by the main CPU. When the focal distance of the lens is great, the period is shortened, and a error in the correction is decreased. On the contrary, when the focal distance is short, the period is made long, and the blurring can be detected even on the object of the low brightness.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention detects a blur of a subject image, drives at least a part of a photographing optical system during film exposure to correct the blur of the subject image, and detects a blur-free photograph. And a camera capable of correction.

[0002]

2. Description of the Related Art A blur detection camera or a blur correction camera using an image sensor has been proposed. The blurring of the subject image changes every moment in its speed and direction, and is particularly remarkable when it is caused by camera shake.
Therefore, in order to fully correct blurring, "blurring detection" →
It is necessary to continuously and repeatedly perform the operation "correction by driving the optical system" throughout the exposure. On the other hand, in the conventional camera for detecting and correcting blur, the blur detection and correction are repeated with a certain period of time.

On the other hand, it is necessary to shorten the repetition cycle of blur detection / correction as the focal length of the taking lens increases. The reason will be briefly described below. FIG. 17 is a diagram showing the relationship between the focal length of the taking lens and the blur. FIG. 17
In (A), the focal length is f, the distance from the subject to the lens is b, and the distance from the lens to the film is a.
Further, the image of the point A on the subject is A '.

FIG. 17B shows a state in which the camera in the state as shown in FIG. 17A has a camera shake of amplitude θ and the position of point A ′ vibrates at an angular frequency ω. When the camera is blurred by θ / 2 in the downward direction in the figure, the point A becomes the position of the point Ad when viewed from the camera, and the point A on the film surface.
Form an image on d '. Further, when the image is blurred by θ / 2 in the upward direction in the figure, the point A becomes the position of the point Au when viewed from the camera,
An image is formed at a point Au ′ on the film surface.

Here, the point of intersection of the optical axis and the film surface is defined as point O, and the x axis extends upward from point O. And point O and point A '
If the distance between them is x and the point A'is vibrating sinusoidally on the film surface, the position of the point A'is represented by ◇ x = atan (θ / 2) ・ sinωt (t is time) ◇. be able to. A graph of this is shown in FIG. In FIG. 18, the intervals of t _i , t _{i + 1} , and t _{i + 2} are the blur detection / correction repetition period ts. It is assumed that a blur between t = t _i and t = t _{i + 1} is detected for the blur as shown in the figure. Next, at t = t _{i + 1} , it is predicted that the blurring will continue at the detected blurring speed thereafter (line 1 in the figure), and the correction optical system is driven to cancel this. At this time t =
The blur correction error is the difference between the actual blur (2 in the figure) and the predicted blur (1 in the figure) between t _{i + 1} and t = t _{i + 2} .

As can be seen from the figure, the shake amplitude atan
The larger (θ / 2), the larger the blur correction error. On the other hand, the distance a from the lens to the film is calculated by the image formation formula ◇ 1 / a + 1 / b = 1 / f, and a = bf / (b-
f) It becomes ◇. In normal shooting, b >> f, so that a≈f, and it can be said that the shake amplitude atan (θ / 2) is proportional to the focal length of the lens. Therefore, the blur correction error increases as the focal length of the taking lens increases.

On the other hand, as can be seen from FIG. 18, the repetition period t _s may be shortened in order to suppress the blur correction error. Therefore, in order to perform proper blur detection / correction, it is necessary to shorten the repetition cycle as the focal length of the taking lens increases.

[0008]

As described above, the longer the focal length of the taking lens, the shorter the cycle of shake detection / correction needs to be shortened. On the other hand, when blur detection is performed using an image sensor, it is necessary to lengthen the repetition cycle in order to detect blur even with a low-luminance subject. This is because it is necessary to increase the integration time of the image sensor for a low-luminance subject.

On the other hand, the conventional camera capable of detecting and correcting blur has a constant repetition period. For this reason, if the repetition period is set to be short so that even a lens having a long focal length can perform appropriate blur correction, there is a problem that blur detection of a low-luminance subject becomes impossible. On the contrary, if the repetition cycle is set to be long so that blur detection can be performed even for a low-luminance subject, there is a problem that a blur correction error increases in a photographing lens having a long focal length.

Therefore, an object of the present invention is to eliminate the above-mentioned drawbacks of the conventional example, to perform appropriate blur detection and correction even with a lens having a long focal length, and to perform blur detection and correction even on a low-luminance object. To provide a camera.

[0011]

In order to achieve such an object, a camera capable of detecting and correcting blur according to the present invention is an image sensor and a calculation for calculating blur data of a subject image based on the output of the image sensor. A camera having a means and a blur correction means for performing blur correction based on the blur data, comprises an input means for inputting focal length information of a photographing lens, and performs blur detection and correction based on the input focal length information. It is characterized in that a cycle determining means for determining a cycle is provided.

[0012]

According to the present invention, the repetition cycle of blur detection and correction is determined based on the focal length of the taking lens. When the focal length of the taking lens is short, blurring can be detected even for a low-luminance subject by lengthening the repetition cycle. On the contrary, when the focal length of the taking lens is long, the blur correction can be performed more accurately by shortening the repetition cycle.

[0013]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a main part of a blur detection camera according to the present invention.

Referring to FIG. 1, a photometric SPD (Sili)
The signal from the con photodiode 2 is input to the photometric circuit 3. The photometric circuit 3 calculates the output of the SPD 2 and outputs the calculation result to the main CPU 1. The taking lens 4 is replaceable, and the diaphragm drive means 5 and the focus adjustment drive means 6 drive the diaphragm and adjust the focus. The photographic lens circuit 7 stores the F-number of the lens unique to the photographic lens 4, the focal length, and various parameters for blur correction. The taking lens circuit 7 further includes an actuator for driving an optical system for blur correction and a control circuit for the actuator.

The AF detection mirror 8 guides a part of the light flux passing through the taking lens 4 to the phase difference type AF detection module 13. The AF detection mirror 8 is retracted by the AF detection mirror driving means 9 so as not to block light to the film during film exposure. Shutter curtain 1 for focal plane shutter
Reference numeral 0 includes a front curtain and a rear curtain, and shutter shutter drive means 11
Driven by. The switch S1 is turned on by the first stroke of the shutter button to perform photometry and AF operation. The switch S2 is turned on by the second stroke of the shutter button. The film winding means 14 winds and rewinds the film. The switch SC is a switch for operating the continuous shooting mode.

The blur detection optical system 18 is an optical system for guiding light to a CCD for blur detection, which will be described later.

Reference numeral 12 denotes camera shake detection / correction means for detecting and correcting camera shake. Reference numeral 12 has a calculator for performing shake calculation and a control CPU for controlling the shake correction optical system. The camera shake detection / correction means 12 will be described in detail later.

The main CPU 1 is further connected to an exposure correction amount input means 16 for performing exposure correction for an exposure amount determined by a photometric value and a film sensitivity setting means 17 for setting film sensitivity.

FIG. 2 is a diagram showing an optical system of the blur detection camera according to the present invention. Referring to FIG. 2, the taking lens is 4
Includes a shake correction lens 20 and a shake correction lens drive device 21 for moving the shake correction lens 20. Since the details of the correction lens driving device 21 are known, the description thereof will be omitted. The pellicle mirror 22 is fixed inside the camera body 32, and a part of the light flux passing through the taking lens 4 is reflected toward the finder optical system, and the rest is transmitted toward the shutter curtain 10. A part of the light flux transmitted through the pellicle mirror 22 is guided to the phase difference AF detection module 24 by the AF detection mirror 8. The AF detection mirror 8 is retracted by the AF detection mirror driving means 9 (not shown) to a position where the light flux during exposure which reaches the film during film exposure is not blocked.

The light beam reflected by the pellicle mirror 22 passes through the condenser lens 25 and is transmitted to the penta prism 2.
Enter 6. A photometric SPD2 is provided on the eyepiece 27 side of the pentaprism 26. One surface 26a of the pentaprism 26 is a half mirror, and a part of the incident light flux is extracted to the outside of the bent prism 26. The extracted light flux passes through the blur detection optical system 18 and is reflected by the reflecting mirror 28 to reach the blur detection sensor 29. The blur detection optical system 18 re-images the image on the finder focal plane 30 on the blur detection sensor 29. The blur detection sensor 29 is a CCD area sensor. The penta prism 26 is not limited to the glass block type and may be a hollow penta mirror having an air layer inside.

As described above, in the present invention, since the light of the subject is transmitted to the area sensor 29 serving as the blur detection sensor by using the velicle mirror 22, the blur detection is performed by the image detection system in the TTL system camera. Further, since the pentaprism 26 is used in the optical path to the blur detection sensor 29, the blur detection sensor 29
It is not necessary to provide a separate optical path to.

FIG. 3 is a view showing a finder screen. A in FIG. 3A is a transparent blur detection region, and the presence of this region secures an optical path to the sensor 29 and increases the amount of light to the sensor 29. The area represented by b is the matte portion of the finder screen, where light is scattered, an image is formed on the matte area as usual, and focus is detected. The area indicated by a is also used for displaying the finder display frame of the blur detection area. FIG. 3B is a diagram showing a case where the entire surface of the finder screen is transparent. A part a near the center of the finder screen b is imaged on the imaging surface by the blur detection optical system 18. This image is used by the blur detection sensor 29.

Referring to FIG. 2, the taking lens 4 is attached to the camera body 32 via the lens mount 31 and can be replaced.

If a normal lens without a camera shake correction lens is attached to the camera body 32,
A warning is displayed on the viewfinder screen when blur is detected. The electronic finder may be warned of the camera shake with the words "Camera warning".

Next, referring to FIG. 4, camera shake detection / correction means 1
The contents of 2 will be described. The camera shake detection sensor 29 includes a CCD image pickup unit 41, an output amplifier 42 that amplifies a CCD output,
An illuminance monitor 43 for controlling the integration time of the CCD,
And a photometric circuit 44. The clock generator 45 is
The output of the photometric circuit 44 is detected to set the integration time of the CCD and the gain of the output amplifier. The clock generator 45 also generates a clock for driving the CCD, a clock for the A / D converter 51, a D / A converter 48, a sensitivity variation correction memory 49, and a dark output correction memory 50. The output of the blur detection sensor 29 passes through the differential amplifier 46 and the gain control amplifier 47, and the A / D converter 5
Input to 1. The dark output correction memory 50 and the sensitivity variation correction memory 49 store data for correcting CCD sensitivity variation and dark output, respectively. The dark output correction memory 50 outputs data to the D / A converter 48. The output signal of the D / A conversion is input to the differential input of the differential amplifier 46. As a result, the dark output of the CCD 41 is corrected. The gain control amplifier 47 is an amplifier whose amplification degree is controlled by a digital signal. The gain control amplifier 47 is a sensitivity variation correction memory 49.
Controlled by the sensitivity variation correction data stored in, the sensitivity variation of the CCD output is corrected.

The output signal of the A / D converter 51 is stored in the standard memory 52 or the reference memory 53. The address generator 54 includes a standard part memory 52 and a reference part memory 5.
The address data necessary for the operation of each memory 3 is generated.

The calculator 55 includes a subtraction circuit 56, an absolute value circuit 57, an addition circuit 58 and a register 59. The data of the standard part memory 52 and the reference part memory 53 are given as inputs.

The calculation result from the calculator 55 is a correlation result memory 60 and a vertical contrast memory 61 depending on the type of calculation.
Alternatively, it is stored in any of the lateral contrast memories 62. These memories are connected to the control CPU 63 and can be accessed from the control CPU 63. The control CPU 63 also controls the address generator 54 and the clock generator 45.

Next, a blur detection method and a blur amount calculation will be described. First, the blur detection sequence will be described. In the present invention, the object image is two-dimensionally detected using the area sensor 29 capable of detecting two-dimensional image data.
The blurring of the image is detected by detecting the time-dependent shift of the subject image using the area sensor 29.

The CCD 41 is an area sensor of I × J pixels. The standard part memory 52 and the reference part memory 53 are I × J word memories, and the correlation result memory 60
Is a memory having a capacity of H × H words. CCD4
Consider the light receiving surface of 1 divided into M × N blocks. Each block is composed of adjacent K × L pixels. The vertical contrast memory 61 and the horizontal contrast memory 62 are M, respectively.
A memory having a capacity of × N words. Hereinafter, the procedure of blur detection will be described. Here, I = 68, J = 52,
It is assumed that K = 8, L = 8, M = 8, N = 6, and H = 5. A /
The resolution of the D converter 51 is 8 bits, the register 59 is 14 bits, and each word of the correlation result memory 60, the vertical contrast memory 61, and the horizontal contrast memory 62 is represented by 14 bits.

FIG. 5 is a schematic view of a part of the light receiving portion of the CCD 41. One small square grid represents a unit pixel. The lower left pixel in the figure is a pixel (1, 1), and the upper right pixel is a pixel (68, 52). Except for the two pixels on the outer circumference of the light receiving portion, the light receiving portion of the CCD 41 is divided into blocks composed of 8 pixels × 8 pixels. The portions surrounded by thick lines in the figure correspond to each block. The lower left block is block (1, 1), and the upper right block is block (8, 6).

The blur detection sequence is roughly divided into three parts: ◇ (1) contrast calculation and block selection ◇ (2) correlation calculation ◇ (3) interpolation calculation ◇

The contrast calculation is a calculation performed for block selection. In each part of the subject imaged on the light receiving part of the CCD 41, there is a part suitable for shake detection and another part not suitable for shake detection. In this embodiment, the contrast of the subject is calculated in order to select a portion suitable for blur detection. The contrast of the object is calculated for each block, and blurring is detected using a total of eight blocks, four blocks having a large vertical contrast and four blocks having a large horizontal contrast.

(1) Description of Contrast Calculation and Block Selection ◇ First, the output of the CCD 41 is stored in both the standard memory 52 and the reference memory 53. The vertical contrast and the horizontal contrast of each block are calculated using this data. The calculation is block (1,1),
Lateral contrasts of (2,1), ..., (7,6), (8,6), blocks (1,1), (2,1) ,.
6) and (8, 6) are performed in the order of vertical contrast.

The output of the pixel (i, j) of the CCD 41 is A
Let (i, j) be the horizontal contrast of block (k, *).

[0036]

[Equation 1]

The vertical contrast

[0038]

[Equation 2]

It is defined as

However, the above * is a symbol added to avoid confusion between the lowercase letter l (el) and the number 1 in the text. Therefore, in the text, * means the same as the cursive ell in drawings and mathematical formulas.

A procedure for executing this calculation formula by the hardware shown in FIG. 4 will be described below. The content of the reference portion memory 52 corresponding to the output of the pixel (i, j) of the CCD 41 is set to R (i,
j), the content of the reference memory 53 is S (i, j).
First, the register 59 is cleared. Next, the subtractor 56
R (i, j) at one input and S (i +
1, j) {however, i = 8 (k-1) +2, j = 8 (*-
1) +2} is given so that the address generator 5
The address is transmitted from the terminal 4. The data subtracted by the subtraction circuit 56 has its absolute value taken by the absolute value circuit 57, added by the addition circuit 58 to the contents of the register 59, and stored in the register 59. Next, the address that should be i = i + 1 is sent from the address generator 54, and the same calculation is performed.
In this way, i = 8 (k-1) +2 to 8k + 2, j
= 8 (*-1) +3 to 8 * ++ 2, the register 59 stores

[0042]

[Equation 3]

Is stored.

Standard part memory 52 and reference part memory 53
The same content A (i, j) is stored in
Since (i, j) = R (i, j) = S (i, j),
The content of register 59 is the horizontal contrast

[0045]

[Equation 4]

It is When one block has been calculated, the content of the register 59 is transferred to and stored in the lateral contrast memory 60 at a position corresponding to that block.

Similarly, the horizontal contrasts of all the remaining blocks are calculated and stored in the horizontal contrast memory 62.

Next, the contrast in the vertical direction is calculated.
The register 59 is first cleared, as in the case where the lateral contrast is calculated. Next, R (i, j) is input to one input of the subtraction circuit 56 and S (i, j + 1) is input to the other input.
{However, i = 8 (k-1) +3, j = 8 (*-1) +2,
The address is sent from the address generator 54 so that The data subtracted by the subtraction circuit 56 has its absolute value taken by the absolute value circuit 57, is added to the contents of the register 59 by the addition circuit 58, and is stored in the register 59. Next, the address which should be i = i + 1 is sent from the address generator 54, and the same calculation is performed. In this way, i = 8 (k-1) +3 to 8k + 2, j = 8
The processing is performed in the range of (* -1) +2 to 8 * ++ 2.
As a result, in the register 59,

[0049]

[Equation 5]

Is stored. Where A (i, j) =
Since R (i, j) = S (i, j), the register 72
Is the vertical contrast

[0051]

[Equation 6]

It is When the calculation for one block is completed, the contents of the register 59 are changed to the vertical contrast memory 61.
The data is transferred to and stored in the place corresponding to the block of the.

Similarly, the vertical contrasts of all the remaining blocks are calculated and stored in the vertical contrast memory 61.

The control CPU selects the block with the highest contrast from the vertical and horizontal contrasts of each block obtained as described above. Next, a block having the highest contrast is selected from the remaining blocks in a direction different from the contrast direction selected first (for example, if the first selection is the vertical direction, the horizontal direction). Similarly, while changing the direction of contrast, eight blocks (4 in each of the vertical and horizontal directions) are selected in the order of blocks with high contrast. B for the selected block
1 to B8.

(2) Description of Correlation Calculation As the contents of the reference part memory 52, the data for which the contrast is calculated is left as it is as the reference image data. New data is read from the CCD 41 one after another, and is written in the reference unit memory 53 as a reference image each time. Each time new data is written in the reference memory 53, the standard image and the reference image are compared, and the spatial shift between the two is detected as a blur of the image. The image blur is obtained by the correlation calculation and the interpolation calculation described below.

Block Bk (k = 1, 2, ... 8) in the following equation
Defines the correlation value of.

[0057]

[Equation 7]

Considering the case of * = m = 0,

[0059]

[Equation 8]

It becomes This value is obtained by adding one block of the absolute value of the difference between the data of the same pixel in the standard image and the reference image. Considering cases other than * = m = 0, C
k (*, m) is obtained by adding the absolute value of the difference between the data of the pixel (i, j) of the standard image and the data of the pixel (i +, j + m) of the reference image for one block. Correlation value obtained by adding the correlation value for 8 blocks

[0061]

[Equation 9]

It is assumed that

The reference section memory 52 is connected to one input of the subtraction circuit 56, and the reference section memory 53 is connected to the other input. Therefore, after clearing the register 59, R (i, j) and the other input S (i ++, j + m) are input to one input of the subtraction circuit 56. Then, the control CP is processed so that data in a predetermined range of i and j is processed.
The address determined by U is sent from the address generator 54. Then, C (*, m) is obtained in the register 59. This is transferred to and stored in a predetermined address of the correlation result memory 60. By changing the values of * and m and repeating the above process, C (*, m):
All (*, m = -2, -1, 0, 1, 2) are obtained.

(3) Description of interpolation calculation When the calculated correlation value C (*, m) is arranged with * on the horizontal axis and m on the vertical axis, an array like a number is obtained.

C (-2,2) C (-1,2) C (0,2) C (1,2) C (2,2) ◇ C (-2,1) C (-1,1) C (0,1) C (1,1) C (2,1) ◇ C (-2,0) C (-1,0) C (0,0) C (1,0) C (2,0 ) ◇ C (-2, -1) C (-1, -1) C (0, -1) C (1, -1) C (2, -1) ◇ C (-2, -2) C ( -1, -2) C (0, -2) C (1, -2) C (2, -2) ◇ If there is no deviation between the standard image and the reference image, C
(0,0) becomes 0, and has a larger value toward the outside of the array. If the reference image is shifted to the right by * ₀ pixels and is shifted upward by m ₀ pixels from the reference image, C
(* ₀ , m ₀ ) becomes 0, and has a larger value as it goes away from it. However, the shift of the image is not always an integral multiple of the pixel. Further, since the speed of image blur is not constant, the blurring of the image due to image blur may differ between the standard image and the reference image. The distribution of the correlation value C (*, m) in such a case is represented by contour lines as shown in FIG. With reference to FIG. 6, the values are actually obtained only on the grid points in the figure, but contour lines are drawn assuming the values between the grids. The contour line has a smaller value toward the center. The center of the contour line is the MP point at the coordinates (x ₀ , y ₀ ), and the reference image is considered to be laterally displaced from the standard image by x ₀ and vertically to y ₀ . Since only the values on the grid points are obtained by the correlation calculation, it is necessary to interpolate the MP points using the data of the grid points. Magnification of shake detection optical system 18, CCD 41
The pixel size and integration time are set so that the MP point is −1.5 <x ₀ , y ₀ <1.5, assuming the actual blur size. The interpolation calculation after completion of the correlation calculation will be specifically described. The calculation in the horizontal direction will be described. First C
Find the minimum C (* ₀ , m ₀ ) of (*, m). * ₀
Values and C (* _0-1, m ₀₎ and _{C (* 0 + 1, m} 0) Figure in magnitude relation between 7 is divided in the case shown in (a) ~ (h) in FIG. 8.

In the case of the condition (a): It is judged that 0 ≦ x ₀ <1. ◇ Point (-1, C (-1, m ₀ )) and point (0, C (0, m)
₀ )) and a point (1, C (1, m ₀ ))
And the * coordinate of the intersection of the line (ii) connecting the points (2, C (2, m ₀ )) is x ₀ .

[0067]

[Equation 10]

In the case of the condition (b): It is judged that 1 <x ₀ <0. ◇ Points (-2, C (-2, m ₀ )) and (-1, C (-1,
m ₀ )) and a point (0, C (0, m
₀ )) and a point (1, C (1, m ₀ ))
Let x _{0 be} the * coordinate of the intersection with i).

[0069]

[Equation 11]

In the case of the condition (c), it is judged that 0 <x ₀ <1. ◇ The following is the same as the case of (a),

[0071]

[Equation 12]

In the case of the condition (d): It is judged that 1 ≦ x ₀ <2. ◇ A straight line (i) connecting points (0, C (0, m ₀ )) and (1, C (1, m ₀ )) and a slope ((C) (2, m ₀ )) Let x _{0 be} the * coordinate of the intersection of i) and the straight line (ii) with the opposite sign.

[0073]

[Equation 13]

In the case of the condition (e), it is judged that -1 ≤ x ₀ <0. ◇ The following is the same as the case of (b)

[0075]

[Equation 14]

In the case of the condition (f), it is judged that -2 <x ₀ <-1. ◇ Point (-1, C (-1, m ₀ )) and point (0, C (0, m)
₀ )) and a line (ii) and a point (-2, C (-2, m
₀ )), and the * coordinate of the intersection of the straight line (ii) with the slope and the straight line (i) with the opposite sign is x0.

[0077]

[Equation 15]

In the case of the conditions (g) and (h): It is determined that the blur amount cannot be detected because the blur exceeding the maximum expected blur occurs.

Although the procedure for obtaining x ₀ has been described above, y ₀ can be similarly obtained.

Next, a method of driving the correction lens 20 shown in FIG. 2 will be described. FIG. 9 is a diagram for explaining a driving method of the correction lens 20. Here, only the lateral component of the image blur is considered. In FIG. 9, the horizontal axis t represents time, and the vertical axis x represents the position of the image. t _-3 , t _-2 , t _-1 , ...
Represents the integration start time of the CCD 41, t _-3 ″,
t ₋₂ ″, t ₋₁ , “... Represents the integration completion time. The integration time TIi is represented by TI _i = t _i ″ −t _i . When the subject is illuminated by an AC light source, the integration time is changed so that the exposure amount of the CCD 41 becomes constant. Is not constant, the integration start time t _-3 ,
t ₋₂ , t ₋₁ , ... Are equal intervals TS. After this TS
Will be called a correction cycle. A solid curve 301 indicates the locus of the image on the CCD 41 when there is image blur.
If blur correction is not performed by the correction lens 20,
The image moves on this curve. Since the blur correction starts from t = t ₀ , the curve 301 is drawn so as to pass through the point (t ₀ , 0). The polygonal line 302 is the locus of the correction lens 20. A broken line 303 represents the locus of the image on the CCD 41 when the correction lens 20 is moved to perform the correction. The points P _-2 , P _-1 , P ₀ , ... Are the integration periods t _{-3 to} t _-3 ″,
_{t -2 ~t -2 ", t -1} ~t -1", indicating an average position of the image in .... CCD data obtained by integrating between t _i ~t _i "is read during _{t i + 1 ~t i + 2} , calculation processing is performed from .t i _{+ 2} position Pi of the image is obtained The correction lens 20 is driven by the obtained data.

In the following description, P _-2 , P _-1 , P
₀ , ... X ₁ , X ₂ , X ₃ , ... X ₁ ', X ₂ ', X ₃ ',
... X ₀ ″, X ₁ ″, X ₃ ″, ... Are used as the name of the point and also as the value of the x coordinate of the point.

In order to make a correction from time t = t ₀ ,
It is necessary to find the speed of blur between t ₀ and t ₁ . The blur speed is calculated from the recent changes in the image positions of two points, and t ₀ ~
Predict that the blurring speed during t ₁ will be the same. t
The latest image position known at the time point of = t ₀ is P ₋₁ . The time from P _-2 to P _-1 is TS- (TI _-3 -T
Since I ₋₂ ) / 2, the predicted value Vx ₀ of the blurring speed from t ₀ to t ₁ is

[0083]

[Equation 16]

It is Therefore, between t ₀ and t ₁ , the correction optical system is driven at the speed of Vx ₀ .

Next, consider the case of time t = t ₁ . At this time, since the correction lens 20 is driven at Vx ₀ , it is moved to the position of X ₀ ″. Since the position of P ₀ is known at this time, the predicted value Vx ₁ of the blur velocity from t _{1 to} t ₂ is calculated. Then it is asked as follows.

[0086]

[Equation 17]

Since the velocity Vx ₁ is obtained based on the latest image position data, it is considered that the predicted value of the blur velocity between t ₀ and t ₁ is more accurate than Vx ₀ . Therefore, the prediction error ERx ₁ is calculated as follows. _{◇ ERx 1 = (Vx 0 -Vx} 1) · TS ◇ position X ₁ of the image from these values t = t ₁ is predicted as follows. ◇ X ₁ = Vx ₀ · TS-G · ERx ₁ ◇ = {Vx ₀ −G (Vx ₀ −Vx ₁ )} · TS ◇ G is called a prediction coefficient, and in FIG. 9, the prediction coefficient G =
The case of 2 is shown. Here, the correction lens 20 needs to be driven to the newly predicted X ₁ point, but driving requires a time of t ₁ to t ₁ ′. Meanwhile, the image is X
Because it is predicted to move to the position ₁ ', the correction lens 2
0 is driven to the position of ⋄X ₁ ′ = X ₁ + Vx ₁ × (t ₁ ′ −t ₁ ) ∘, and is driven at the speed of Vx ₁ between t ₁ ′ and t ₂ .

Next, consider the case of time t = t ₂ . At this time, the correction lens is moved to a position represented by X ₁ ″ = X ₁ + Vx ₁ · TS. At this point, since the position of P ₁ is known, the blur velocity prediction between t _{2 and} t ₃ is predicted. The value Vx ₂ can be calculated as follows.

[0089]

[Equation 18]

It is considered that Vx ₂ is more accurate than Vx ₁ as a predicted value of the blurring speed between t _{1 and} t ₂ . Therefore, the prediction error ERx ₂ is calculated as follows. _{◇ ERx 2 = (Vx 1 -Vx} 2) · TS ◇ position X ₂ of the image from these values t = t ₂ are predicted as follows. ◇ X ₂ = X ₁ ″ -G · ERx ₂ ◇ = X ₁ + {Vx ₁ −G (Vx ₁ −Vx ₂ )} · TS ◇ Here, the correction lens 20 is driven to the newly predicted position X ₂ . it is necessary, but the drive 'is required time. during which the image is X _2' t ₂ ~t ₂ correcting lens 20 since it predicted to move to the position of _{◇ X 2 '= X 2 +} Vx 2 · (T ₂ '-t ₂ ) ◇ is driven, and during the period from t ₂ ' to t ₃ , it is driven at the speed of Vx ₂ .

Next, consider the case of time t = t ₃ . The time correction lens X ₂ "is ◇ X _2" is in the position represented by _{= X 2 + Vx 2 · TS} ◇. Predicted value Vx ₃ shift speed between t ₃ ~t ₄ because you know the position of P2 at this time
Can be calculated as follows.

[0092]

[Formula 19]

[0093] Vx ₃ obtains the prediction error ER ₃ because it is considered more accurate than Vx ₂ as the predicted value of the shake speed between t ₂ ~t ₃ as follows _{◇ ER 3 = (Vx 2 -Vx} 3 ) · TS ◇ position X ₃ of the image at t = t ₃ these values _{◇ X 3 = X 2 "-G} · ERx 3 ◇ = X 2 + {Vx 2 -G (Vx 2 -Vx 3)}・ TS ◇ can be predicted, where the correction lens 20 needs to be driven to the newly predicted position X ₃ , but the driving is from t ₃ to.
Time between t ₃ 'is needed. Since it can be predicted that the image moves to the position of X ₃ ″ during that time, the correction lens 20 is driven to the position of ◇ X ₃ ′ = X ₃ + Vx ₃ · (t ₃ ′ −t ₃ ) ◇, and t ₃ ′ to t _3. Between ₄ is driven at the speed of Vx ₃ .

Thereafter, similarly, X ₄ , Vx ₄ , X ₅ ,
Vx ₅ , ... Is obtained, and the correction optical system is driven.

In this embodiment, the prediction coefficient G is constant. However, the value of G may be changed during the correction sequence depending on the situation of blur detection. The prediction coefficient G may be increased if the prediction error ER does not become small or the sign of the prediction error ER _i does not reverse even after the shake correction is performed several times. Further, when the sign of the prediction error ER _i constantly repeats inversion, the value of G may be reduced.

Referring to FIG. 9, the time TR from time t ₀ to t _z is the photographing exposure time of the camera. As can be seen from FIG. 9, the camera shake correction sequence is repeated many times within the photographing exposure time TR. Therefore,
The correction is not performed only once or twice, but is performed at least several times. In order to carry out blurring correction on a low-luminance object in such a correction method, it is preferable that the correction cycle TS be long and the integration time of the CCD 41 be long. However, if TS is too large, a sufficient correction effect cannot be obtained for fast blur. As mentioned before, it is necessary to shorten the TS for a shooting lens with a long focal length, but according to experiments, TS = 1 for a 50 mm lens.
0ms, T of about 2.5ms with a 200mm lens
The result was that S was suitable.

As described above, t _{i to} t _{i + i} in FIG.
Is TS, and the integration time t of the image sensor CCD 41
_i ~t _i "can not be longer than this to take as long as possible the integration time so that it can cope with a low luminance object is. In this example, changing the TS in accordance with the focal length of the taking lens. Specifically, TS = 1 for focal lengths less than 50 mm
0 ms, TS = 5 ms for 50 mm or more and less than 200 mm,
For 200 mm or more, TS is set to 2.5 ms. In this way, the image sensor CC is
By determining the maximum value of the integration time of D41, a low-luminance subject is dealt with.

As described above, in the present invention, blur detection is performed using the output of the area sensor that detects the subject image, and the blur correction as described above is performed using the detected blur amount.

Next, the interface between the interchangeable lens 4 and the camera body 32 will be described. From the interchangeable lens 4, as information peculiar to the lens, correction magnifications KBLX and KBLY (values corresponding to respective focal lengths when zooming is performed): (image moving amount of film surface) / (driving pulse) and focal length f _L is input.

Then, drive pulse x _i = image movement amount X ₀ / correction magnification KBLX or yi = image movement amount Y ₀ / correction magnification K
Required by BLY. Next, data necessary to drive the above-mentioned lens ◇ 1 direction X: positive and negative 2 drive speed V _xi ◇ 3 drive pulse X _i 4 drive pulse X _i '◇ 5 direction Y: positive and negative 6 drive speed V _yi ◇ 7 drive pulse Y 8 drive pulse Y _i '◇ 9 Home return ◇ is output to the lens side. Here, the origin return signal 9 is set only when the lens is returned to the origin, and the lens side returns the correction lens 20 to the origin only when there is this signal.

In the control on the lens side, the correction lens 20 is driven as described above based on the input data. When undetectable data is output, the same data as the previous data is output.

Next, the control of the integration time of the CCD 41 will be described. FIG. 10A shows a circuit for controlling the integration time of the CCD 41, which includes an illuminance monitor SPD and a photometric circuit. The photometric circuit includes an integrating circuit and a reset circuit. The illuminance monitor SPD is formed near the photographing area of the CCD 41. The photocurrent output from the illuminance monitor SPD is
It is integrated by the integrating circuit. The output of the integrating circuit is proportional to the amount of light incident on the illuminance monitor during the integration period. The illuminance monitor and the photometric circuit are used to keep the output of the CCD 41 constant even when the subject is illuminated by an AC light source such as a fluorescent lamp. This is because if the integration time of the CCD 41 is made constant when the subject under the illumination of the AC light source is imaged, the read output for each time changes. CC
In order to stabilize the read output of D41, the integration time needs to be adjusted according to the illuminance of the subject. In this embodiment, the integration of the photometric circuit is started at the same time when the integration of the CCD 41 is started, and the exposure amount of the CCD 41 is monitored by the output of the integration circuit.

FIG. 10B is a modification of FIG. 10A.

Next, the operation of the integration time control circuit shown in FIGS. 10A and 10B when the CCD 41 is used as a blur detection sensor will be described. The reset circuit is turned on and the integration circuit is reset. The charge in the CCD storage is cleared. When the integration of the CCD 41 is started, the reset circuit is turned off and the photometric circuit starts the integration.

FIG. 11 is a diagram showing the temporal change of the photometric circuit output. The horizontal axis t represents the integration time, and the vertical axis I represents the magnitude of the output of the photometric circuit. When the subject is illuminated by the AC light source, the output of the photometric circuit rises in a curved line as shown in the figure. Set the photometric output to an appropriate reference value I
Compared with 0, when the photometric output I becomes I0, the integration circuit is reset and the integration of the CCD 41 is terminated. C
The output of the CD 41 is read and it is judged whether the exposure amount is appropriate. When the exposure amount is appropriate, the reference value I0 is used for the subsequent exposures. If the exposure amount is inappropriate, for example, the exposure amount is multiplied by k to set I0 × k as a new reference value I0, and the subsequent exposure is performed.

FIG. 12 is a schematic diagram of the chip of the camera shake detection sensor 29.

Reference numeral 210 denotes a sensor chip, and reference numerals 211 to 214 denote pixel columns, each of which has 52 pixels.

There are 68 pixel columns, and the total number of pixels is 68 × 5.
It is 2. Reference numerals 215 to 218 are vertical transfer CCDs, 219 is a horizontal transfer CCD, and 220 is an output amplifier including a charge detection unit. An illuminance monitor SPD 221 is arranged so as to surround the CCD light receiving area so that the illuminance of the CCD light receiving area can be measured as accurately as possible. 222 is a photometric circuit.

The camera sequence including the blur detection and correction described above will be described with reference to the flowcharts of the main CPU and the blur correction control CPU.

First, the sequence of the main CPU 1 will be described. With reference to FIG. 13, the main CPU is in a standby state in a loop of step # 5 (hereinafter abbreviated as step). That is, the loop of # 5 waits for the switch S1 to be turned on by the first stroke of the release button. When the switch S1 is turned on, the AF completion flag AFEF and the signal are reset (# 10), and the necessary circuits such as the photometric circuit 3, the AF detection module 13, the camera shake correction controller 15 are turned on (# 15).
Next, the origin return signal of the correction lens 20 is output to the taking lens 4 (# 20), the timer is reset and then started (# 25), and the lens data of the taking lens 4 is input. After the lens data is input, a correction cycle TS suitable for the focal length f _L of the taking lens 4 is determined (# 32), and open metering is performed (# 35). Note that focal length f _L and TS
As already mentioned, TS = TS for f _L <50 mm
TS = 5 m for 10 ms, 50 mm ≦ f _L <200 mm
When s, 200 mm ≦ f _L , TS = 2.5 ms is determined.

At # 40, it is determined whether or not the AF completion flag AFEF is 1. If it is 1, the program jumps to # 65, and if it is not 1, the program proceeds to # 45. #
At 45, AF detection is performed. In # 50, it is determined whether or not focus is achieved as a result of the AF detection in # 45. If in focus, the program goes to # 55, and if not in focus, branches to # 60. In # 60, after the taking lens 4 is driven to the focus position, the program jumps to # 45. The AF completion flag AFEF is set to 1 in # 55. In # 65, based on the photometric value obtained in # 35, the film sensitivity, and the distance information obtained as a result of AF in # 45, A
E operation is performed. At # 70, the exposure amount of the CCD 41 is set.

At # 75, it is checked whether the shutter stroke second stroke switch S2 is turned on. If the switch S2 is ON, the program is # 1
Go to 05, and if not ON, branch to # 80. # 80
Then, it is determined whether the switch S1 is ON. If the switch S1 is ON, the program jumps to # 25. If it is not ON, the shutter button has not been pressed at all, so at # 85, a timer checks whether or not a fixed time has elapsed. If the fixed time has elapsed, the power of the circuits such as the photometry circuit 3, the AF detection module 13, the camera shake correction controller 15 is turned off (# 100), and then # 5.
The program jumps to the standby state of. If the fixed time has not elapsed, the AF completion flag AFEF is set to 0 (# 90), and the program proceeds to # 80.

At # 105, since the second stroke of the shutter button is being pressed, the release signal is set to H level, and the control CPU is notified of that fact. # 11
After 0, the aperture of the taking lens 4 is narrowed down to the aperture value obtained by the AE calculation of # 65 (# 110). The blur detection signal is H
The level is set, and the shake detection sequence of the control CPU is started (# 120). The AF mirror 8 is retracted (# 125), and the aperture metering is performed (# 13).
0), the shutter speed after the AE calculation is corrected (# 135).

At step # 140, the control CPU waits until the exposure permission signal from the control CPU becomes H level. When the exposure permission signal becomes H level, the shutter curtain 10 is moved and exposure control is performed (# 145). When the exposure is completed, the origin return signal is output to the taking lens 4 (# 150), the correction lens 20 is returned to the origin position, the blur detection signal is set to the L level, and the control CPU is informed that the exposure is completed. (# 155).

The film is wound (# 16
0), it is determined whether or not the continuous shooting mode is set (# 165).
If it is not the continuous shooting mode, the program branches to # 180. If it is the continuous shooting mode, the program proceeds to # 170 and the switch S2 is turned off.
It is determined whether it is N or not. Switch S2 is OFF here
If so, the program branches to # 180 because continuous shooting mode does not perform continuous shooting. If the switch S2 is ON, continuous shooting is performed, so the blur detection signal is set to H level (# 175), and the program jumps to # 130.

After # 180, the AF mirror 8 is returned to the detection position (# 180), the diaphragm of the taking lens 4 is opened (# 185), and the release signal is set to the L level (# 1).
90), the control CPU 63 is informed accordingly. Waiting for S2 to turn off (# 205), the program jumps to # 80 to prepare for the next shooting.

Next, the operation of the control CPU will be described with reference to FIG. At # 15 in FIG. 13, the image stabilization controller 15 is powered on and the sequence is started. Flag and output signal are reset (#B
5) After the integration of the CCD 41 is reset, the integration is started (# B10). In # B30, the release signal from the main CPU is checked, and if it is H level, the program proceeds to # B35, and if it is L level, the program proceeds to # B10. In # B35, the camera waits until the shake detection signal from the main CPU goes high and the shake detection is started.

When the blur detection signal becomes H level, the correction cycle timer is reset (# B40), the integration of the CCD 41 is reset, and then the integration is started (# B42). # B4
At 5, the completion of integration of the CCD 41 is awaited. When the integration is not completed, it is checked at # B47 whether the integration time exceeds TS. When it is over, proceed to # B73, and when it is not over, proceed to # B45. After the integration of the CCD 41 is completed, the read data is dumped to the standard memory 52 and the reference memory 53 (# B50). After the correction period timer has finished counting (# B51), the timer is reset and started (# B52). CCD for # B55
After the integration reset of 41, integration is started.

The contrast calculation is performed (# B6
0), the block used for blur detection is selected (# B65). In # B70, the completion of integration of the CCD 41 is awaited. If the integration is not completed, the process proceeds to # B72 to check whether the integration time exceeds TS. When it is over, proceed to # B73, and when it is not over, #B
Proceed to 70. This # B73 is a process when a sufficient exposure amount cannot be obtained even if the CCD integration time exceeds TS. In this step, it waits until the shake detection signal from the main CPU becomes L level. In this case, blur correction is not performed. After reaching the L level, the process proceeds to # B130.

After the integration is completed, the read data is dumped in the reference memory 53 (# B75). After the correction cycle timer has finished counting in # B76, the timer is reset and started in # B77. In # B80, the integration is reset after the CCD 41 is reset.

Correlation calculation (# B85) and interpolation calculation (#
B90) is performed and the image blur amount is detected. The lens data of the taking lens 4 is read (# B95), and the correction amount and direction of the correction lens 20 are calculated (# B100).

In # B105, it is determined whether or not the blur detection signal from the main CPU is at the L level. If the blur detection signal is at the L level, the blur detection is terminated, and the program branches to # B130. If the blur detection signal is not at L level in # B105,
Since the shake detection is continued, the program proceeds to # B110.
The exposure permission signal is set to the H level (# B110), the main CPU is informed that the exposure may be started, and the lens correction data is output to the taking lens 4 (# B12).
0), the program jumps to # B70. # B13
At 0, it is determined whether the release signal from the main CPU is at L level. If it is L level, the program jumps to # B10 because the shooting is finished. If it is not L level, it is continuous shooting, so the program is # B13.
5 and waits until the detection signal becomes H level. When the detection signal becomes H level, the program jumps to # B40.

FIG. 15 shows a camera shake detection sensor chip according to another embodiment (referred to as a second embodiment) of the present invention. This sensor is composed of four frame interline CCDs. One of the CCDs will be described. Two
24 to 227 are pixel columns, 12 columns in total. 1 for each row
The image sensor is composed of 2 pixels and has a size of 12 × 12 pixels.

Reference numerals 228 to 231 are vertical transfer CCDs and 232.
Is a charge storage unit. 233 is a horizontal transfer CCD, 234
Is an output amplifier including a charge detector. An illuminance monitor 235 is arranged so as to surround the CCD light receiving area so that the illuminance of the CCD light receiving area can be as accurate as possible. Reference numeral 236 is a photometric circuit.

Since the method of detecting blur is basically the same as that of the first embodiment, the different points will be described. The second embodiment has four blocks, and each block has an illuminance monitor and photometric circuit. In the first embodiment, eight blocks are selected from the 48 blocks for blur detection, but in the present embodiment, the four blocks are divided into two blocks for the vertical direction and two blocks for the horizontal direction. .. Then, the correlation calculation and the interpolation calculation may be performed as in the first embodiment.

Next, the integral control method will be described. In the first embodiment, one illuminance monitor and photometric circuit are provided for all blocks, but in the present embodiment, one illuminance monitor and photometric circuit is provided for each block. FIG. 16 is a block diagram of the camera shake detection sensor in the second embodiment. The integral control will be briefly described with reference to this figure.
The photometric circuit of each block outputs the photometric data of each block to the integration control unit in the clock generator. The integration control unit A / D converts the pixel data of each block every time the integration of each block is completed based on the photometric data.
Convert and transfer to the appropriate part of memory. The integration controller also shifts the charge accumulated in the pixel to the CCD transfer unit,
A pulse (shift pulse) for ending the integration is output. This shift pulse can be input to each block independently. This is to end the integration even while the other blocks are being transferred. Further, it is possible to transfer charges and output pixel data only in a desired block by the block selection signal and the multiplexer. Note that photometric reset is performed simultaneously for all blocks.

As described above, in the second embodiment, the detection sensor is divided into four blocks, and the integration control is performed independently of each other. Therefore, more suitable integration control can be performed for each block.

In the first and second embodiments, the lens interchangeable camera has been described. The present invention can be applied not only to cameras with interchangeable lenses, but also to cameras without interchangeable lenses. More specifically, in a camera having a lens with a variable focal length, that is, a so-called zoom lens, the focal length of the lens may be detected at any time, and the repetition cycle of blur detection / correction may be determined based on the focal length information. ..

[0129]

As described above, the camera of the present invention is
The cycle of blur detection and correction is determined based on the focal length information of the taking lens. Therefore, blur detection / correction is performed at a cycle suitable for the focal length of the taking lens. As a result, even in a lens having a long focal length, it is possible to prevent the blurring correction error from increasing and perform more accurate correction. For a lens with a short focal length, C
The CD integration time can be made long, and blurring detection / correction can be performed even on a subject having lower brightness.

[Brief description of drawings]

FIG. 1 is a block diagram showing a main part of a camera of the present invention.

FIG. 2 is a sectional view showing an optical system of the camera of the present invention.

FIG. 3 is a diagram showing a viewfinder screen of the camera of the present invention.

FIG. 4 is a block diagram showing the contents of camera shake detection and correction means.

FIG. 5 is a schematic view of a part of a light receiving portion of a CCD.

FIG. 6 is a graph showing the distribution of correlation values by contour lines.

FIG. 7 is a graph showing a calculation method of interpolation calculation.

FIG. 8 is a graph showing a calculation method of interpolation calculation.

FIG. 9 is a graph showing a driving state of a correction lens.

FIG. 10 is a circuit diagram of a circuit that controls integration time.

FIG. 11 is a graph showing a temporal change in the output of the photometric circuit.

FIG. 12 is a schematic view of a camera shake detection sensor chip.

FIG. 13 is a flowchart showing an operation sequence of the main CPU.

FIG. 14 is a flowchart showing an operation sequence of the control CPU.

FIG. 15 is a schematic view of a chip of a camera shake detection sensor according to another embodiment.

FIG. 16 is a block diagram of a camera shake detection sensor according to another embodiment.

FIG. 17 is a diagram showing the relationship between the focal length of the taking lens and the blur.

FIG. 18 is a graph showing a blur state when an image on a film surface vibrates sinusoidally.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Main CPU 7 Shooting lens circuit 12 Camera shake detection / correction means 20 Camera shake correction lens 21 Camera shake correction lens drive 29 Camera shake detection sensor 41 CCD 55 Computing unit 63 Control CPU 210 Camera shake detection sensor chip

Claims (2)

[Claims] 1. An image sensor for detecting blurring, and based on imaging data output by the image sensor,
A calculation unit that calculates the blur data of the subject image, a blur correction unit that drives a part of the photographing optical system of the camera to correct the blur based on the blur data output from the calculation unit, and the photographing lens A camera capable of blur detection and correction, comprising: input means for inputting focal length information; and cycle determining means for determining a cycle of blur detection and correction based on the focal length information. 2. The blur detection and correction according to claim 1, wherein the cycle determining means shortens the cycle as the focal length of the photographing lens is longer, and conversely lengthens the cycle as the focal length is shorter. Possible camera.