JPH0670217A - Apparatus and method for correction of image - Google Patents

Abstract

PURPOSE: To make it possible to correct the swing of an image due to a hand shake, to increase the processing speed and to simplify the hardware by detecting a movement vector only of a stop image and making the correction. CONSTITUTION: The output of an image pickup element 31 is converted 32 from analog to digital into image data. A stop image section determining circuit 33 detects the section of a stop image from the image data. A control part 38 controls the image area in a specific section of a memory 34 according to a section address from the circuit 33, stores data representative points of the stop image section in a section memory 35 to provide for the detection of a movement vector of a next frame. A movement vector detecting circuit 36 calculates the image data and the representative points of the stop image section in the memory 35 to obtain the movement vector. An integrator 37 accumulates the movement vector. The control part 38 outputs data of the memory 34 from an address moved equally to the data from the integrator 37. The output data become a stable image which is corrected up to the accumulated value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal processing technique, and more particularly to an image correction apparatus and method for obtaining an accurate image by a camera recorder or a camera. Recently, due to the increasing demand for camera recorders, development of technologies for downsizing, weight reduction, and high-magnification zoom has been actively pursued. Along with this, a technique for obtaining a stable image is required.

[0002]

2. Description of the Related Art FIG. 1 is a diagram showing an image correction device of a general camera recorder. The image pickup device (CCD) 11 converts an optical signal from a subject into an electric analog signal, and an A / D converter 12 converts the analog signal into a digital signal. This digital signal is stored in the frame memory 18 and further stored in the representative point memory 13 for calculation in a frame. Representative point memory 1
Reference numeral 3 is for storing the representative points of the previous frame, and the frame memory 18 is for storing an image of one frame.

The motion vector detector 14 is an A / D converter 1.
The motion vector for each area is detected from the output signal of 2 and the representative point of the previous frame of the representative point memory 13. Here, the motion vector expresses the amount of movement of the image between two frames before and after by a horizontal vector and a vertical vector, and the representative point is a fixed area of a fixed interval selected to reduce the number of calculations. It is replaced with dots. (FIG. 2 shows the case where the motion vector is divided into four areas.) As a typical method for searching the motion vector, there is a band extraction representative point matching technique.
This band extraction representative point matching technique uses a representative point method by removing a high frequency band having a lot of noise components by using a band pass filter in order to reduce errors caused by noise and improve the S / N ratio. It is to reduce the amount of calculation and the size of hardware.

The integrator 15 adds the motion vector detected by the motion vector detector 14 to the motion vectors of each frame accumulated during that period to generate the following new accumulated value. CO = K * COp + V Here, CO is a cumulative value up to the current frame, K is a cumulative constant, COp is a cumulative value up to the previous frame, and V is a motion vector.

The interpolation control unit 17 generates a control signal according to the accumulated value to control the frame memory 18 and the interpolator. The interpolator 19 performs a function of filling a defect of a pixel that occurs when the image is enlarged (zoomed) by a control signal of the interpolation controller 17 in proportion to the distance from the brightness of an adjacent pixel or the like. Reference numeral 16 is an intermediate value controller, which is used to obtain all motion vectors from the motion vectors of each area, and the intermediate value of each motion vector becomes the motion vector of all frames.

On the other hand, the entire screen is displayed as shown in FIG.
When divided into two areas 23, 24, 25, 26, the motion vector for each area 23, 24, 25, 26 can be obtained from the following equation. M (i, j) = Σ | S (x + i, y + j) -Sp (x, y) | (where x = i, y = j) where S is the input data of the current frame and Sp is the previous It is the input data of the frame, (x, y) is the position of the representative point, and (x + i, y + j) is the position moved about (i, j) from the representative point.

The motion vector becomes (i, j) when M (i, j) has the minimum value. Reference numeral 21 indicates a representative point,
Reference numeral 23 shows an output screen.

[0008]

However, such a conventional technique has the following various problems. That is, at first, the motion vector of the moving image has no correlation with camera shake, and the influence of the motion vector of the moving image cannot be completely eliminated by the band extraction representative point method. Therefore, the calculated motion vector has a difference from the motion vector due to shaking.

Second, when the moving image occupies a large portion, the moving vector motion vector has a greater effect than the moving vector motion vector. Therefore, the obtained motion vector becomes a value closer to the motion vector of the moving image which does not care about the motion vector of the hand shake. In order to solve this, the fuzzy function is introduced into the microprocessor, and the motion vector of any area regarded as the motion vector by the moving image is
Although it can be excluded from the process of obtaining the whole motion vector, the obtained whole motion vector becomes uncertain when there are many areas which are regarded as the motion vector by the moving image.

Third, since the size of each area is large and the number of representative points is large, the storage location is large and the number of operations is large. Therefore, the processing speed decreases and the hardware becomes complicated. An object of the present invention is to provide a technique for correcting the problem of image shaking caused by hand shaking.

Another object of the present invention is to provide a technique for detecting image shake of hand shake in a minimum area.

[0012]

According to the present invention, a stop image area determination circuit for determining an area of a stop image from input image data, a memory for storing image data of a predetermined input section, A memory for storing the area data of the still image of the input image data, a detection circuit for detecting the motion vector of the still image based on the image data and the area data of the still image, and a cumulative value by accumulating the motion vectors. The image data memory is controlled by the integrator for creating the image data and the cumulative value and the area data of the stopped image, and the image data that is similarly corrected to the cumulative value is output from the image data memory. This is achieved by an image correction device having a controller that causes the area data of the still image to be output.

[0013]

Embodiments of the present invention will now be described in detail with reference to the accompanying drawings. FIG. 3 is a block diagram of an image correction apparatus according to an embodiment of the present invention. The configuration is such that the optical signal of the subject is converted into an electrical analog signal and the image sensor 3 is used.
1, an A / D converter 32 for converting an electrical analog signal into a digital signal, and a still image area determination for determining an area of a still image along with image data digitally converted by the A / D converter 32. Image data of a predetermined section digitally converted by the circuit 33 and the A / D converter 32, for example, one frame of image data or one filled image data (from the present embodiment, the case of one frame of image data will be described. Memory 34 for storing the still image) and an area memory 3 for storing the still image area of the image data of the previous frame previously determined by the still image area determination circuit 33.
5, a motion vector detection circuit 36 for detecting a motion vector from the image data digitally converted by the A / D converter 32 and the image data stored in the area memory 35, and the motion detected by the motion vector detection circuit 36. An integrator 37 for adding a vector to the motion vectors of each frame accumulated during that period to generate a new accumulated value
And controlling the frame memory 34 according to the accumulated value accumulated by the integrator 37 and the image data of the area determined by the stop image area determination circuit 33, and determines the stop image area data determined by the stop image area determination circuit 33. The control unit 38 is configured to store the image in the memory 35 and output the screen corrected to the cumulative value.

The still image area determination circuit 33 detects the area of the still image from the image data digitally converted by the A / D converter 32. Blurring development is used to distinguish between a moving image and a still image. Blurring development refers to development in which the contour portion of an image becomes dark due to the movement of the subject during the imaging time. Therefore, a portion having a clear contour can be regarded as a still image. In order to store the image data of the area determined by the stationary image, the address (i, j) is used to address (i + L-1, j + N-).
The L * N number of data up to 1) is stored in the area memory 35. According to the embodiment of the present invention, as shown in the following equation, the contour of the image is obtained, and the area having the largest contour is determined as the area of the still image.

M (i, j) = ΣΣed (x, y) (where x = i to ∧, y = j to N) ed (x, y) = √ {(S _{x, y} −S _{x, y +1} ) ² + (S _{x, y} −S _{x + 1, y} ) ² } where ed (x, y) is the size of the contour at the (x, y) point, and S _{x, y} is It is the brightness of the pixel at point (x, y), L is the lateral size of the area, and N is the vertical size of the area.

On the other hand, ed (x, y) can be expressed by the following equation. ed (x, y) max (| _{Sx, y- Sx, y + 1} |, | _{Sx, y- Sx + 1, y} |) (i, j) = max (M _{(i, j)} ), (I, j) is determined by the address of the still image area. The control unit 38 controls the image area of the predetermined section stored in the memory 34 by the area address received from the stopped image area circuit 33, stores the data representative point of the stopped image area in the area memory 35, and Be prepared for detection of frame motion vectors.

The motion vector detection circuit 36 calculates the motion vector by computing the image data digitally converted by the A / D converter 32 and the representative point of the still image area stored in the area memory 35. As described above, the band extraction representative point matching technique can also be used. The integrator 37 accumulates the motion vectors detected by the motion vector detection circuit 36 to obtain a new motion vector cumulative value (CO = K
* COp + V).

The controller 38 causes the data in the memory 34 to be output from the address moved to the data received from the integrator 37. This output data becomes a stable image corrected to the cumulative value. Further, the data up to the address corresponding to the size of the still image area is output from the frame memory 18 to the area memory 52 according to the area address detected by the still image area determining circuit 33 and stored.

According to the present invention, a moving image and a still image are separated, a motion vector of only the still image is obtained, and the motion vector of the still image is regarded as a motion vector due to shaking. That is, as shown in FIG. 4, the motion vector of the moving image and the motion vector of the stationary image have no correlation with each other, and the motion vector of the stationary image is generated by the hand shake.
Further, as shown in FIG. 5, the stop image area is determined by finding the contour of the image (step S1), finding the contour fit for each area (step S2), and stopping the area having the largest contour fit. The image area is determined (step S
3).

As described above, according to the present invention, since the part of the moving image is excluded from the motion vector detection, the error result due to the influence of the motion vector of the moving image can be eliminated. In addition, it is possible to obtain an accurate motion vector without the need for an additional function (for example, fuzzy) to be introduced in order to reduce the influence of the moving image. Also,
Since the capacity of the memory for detecting the motion is small and the number of representative points is small, the number of calculations is reduced, thereby increasing the processing speed and further simplifying the hardware.

[Brief description of drawings]

FIG. 1 is an image correction circuit diagram of a general camera recorder.

FIG. 2 is a view showing an example in which a screen is divided into four areas.

FIG. 3 is a block configuration surface of an image correction apparatus according to the present invention.

FIG. 4 is a diagram for explaining a method of detecting hand shake due to a change between a moving image and a still image in frame data.

FIG. 5 is a diagram illustrating a method of determining a still image.

[Explanation of symbols]

 31 ... Image sensor 32 ... A / D converter 33 ... Stop image determination circuit 34 ... Memory 35 ... Area memory 36 ... Motion vector detection circuit 37 ... Integrator 38 ... Control unit

Claims (7)

[Claims] 1. A still image area determining means for determining an area of a still image from input image data, a storage means for storing image data of the input predetermined area, and the input image data. Storage means for storing the area data of the still image, motion vector detection means for detecting a motion vector of the still image along the area data of the image data and the still image, and accumulation of the motion vector Integrating means for making a cumulative value by controlling the image data storage means along the cumulative value and the area data of the still image, from the image data storage means of the image data corrected in the same manner as the cumulative value. The image is characterized by comprising control means for outputting and for outputting the area data of the still image to the storage means for the area data of the still image. Correction device. 2. The image correction apparatus according to claim 1, wherein the area of the still image is an area having a maximum contour matching of the images. 3. The image correction apparatus according to claim 2, wherein the area of the still image is determined by the following equation. M (i, j) = ΣΣed (x, y) (where x = i, y = j) ed (x, y) = √ {(S _{x, y} −S _{x, y + 1} ) ² + (S _{x, y−} S _{x + 1, y} ) ² } where M (i, j) is a still image, an area, and ed
(X, y) is the size of the contour at the (x, y) point, and S
_{x, y} is the brightness of the pixel at the point (x, y), L is the horizontal size of the area, and N is the vertical size of the area. 4. The image correction apparatus according to claim 1, wherein the area data of the still image is a representative point of the area. 5. The image correction apparatus according to claim 1, wherein the image data of the predetermined section is image data of one frame. 6. The image correction apparatus according to claim 1, wherein the image data of the predetermined section is image data of one field. 7. A still image area of an image correction device comprising the steps of calculating the size of the contour of each area for a number of predetermined areas of an image and detecting the area where the sum of said contours is the largest. How to decide.