JPH05137048A - Video camera jiggle correction circuit - Google Patents

Abstract

PURPOSE:To prevent the misdetection of the video camera jiggle caused by the transient characteristic of a horizontal or vertical filter. CONSTITUTION:The digitization outputs of the input video signals obtained in a photographing area are successively inputted to a horizontal filter 2 and a vertical filter 3 and then inputted to a key point storage means 5 and a block- based correlation detector means 6 with band limitation. The means 5 stores the filter output at a fixed key point corresponding to a detection area which displaces in the areas except the transient areas of both filters 2 and 3. The meant 6 compares the filter output of each detection area and the storage data on a fixed point to obtain the difference absolute value. These absolute values are cumulatively added together at the same position of the detection area in a common detection block. Thus a block-based correction vector is obtained. Based on this vector, a differential vector is obtained. Then the position of the detection area is set based on the differential vector for the next field.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a camera shake detection circuit for a video camera.

[0002]

2. Description of the Related Art A video camera that detects a motion vector by a representative point matching method and corrects the camera shake based on the camera shake detection data in the motion vector is, for example,
NatinalTechnical Report Vol.37 No.3 Jun.1991 P
48-54.

In this representative point matching method, the video signal level of one field before at a plurality of fixed representative points and the video signal level of the sampling point in the detection area including the representative point are compared, and the difference is the largest. A sampling point of the current field with a small degree of correlation is obtained, and the difference between the sampling point and the representative point is specified as the motion of the subject, that is, the motion vector.

In order to specify the motion vector with higher accuracy by the representative point matching method, it is effective to provide a filter output in which the spatial frequency in the horizontal direction or the vertical direction is band-limited in the preceding stage of the motion vector detection circuit. Is also known.

[0005]

However, when the video signal is filtered by the representative point matching method in order to improve the detection accuracy of the motion vector, the filter output level at the video start portion immediately after the blanking period ends is the transient of the filter. It was found to be disturbed by the response characteristics.

Therefore, it is necessary to avoid the detection of the motion vector by the representative point matching method in the filter transition area of the imaging area where the transient response characteristics of the filter reach.

[0007]

Therefore, the present invention is characterized in that the motion detection area constituted by the detection area is set outside the filter transient area where the filter means exhibits a transient response characteristic.

[0008]

Thus, according to the present invention, motion detection is performed outside the filter transient region.

[0009]

The present invention will be described below with reference to the illustrated embodiments.

In this embodiment, the motion detection is executed based on the output through the filters which limit the spatial frequency in the vertical direction and the horizontal direction, respectively. In this embodiment, the motion vector is detected by the representative point matching method, and four detection blocks are provided in the image pickup area as shown in FIG. Each detection block is divided into a total of 16 detection areas, which are vertically and horizontally divided into four equal parts, and a representative point is determined corresponding to each detection area. Within each detection area corresponding to the video signal level of each representative point one field before. The video signal level of the current field at the sampling point of is compared and the absolute value of the difference is obtained as correlation data. For each detection area of the same detection block, absolute value data of sampling points having the same positional relationship with the representative point are cumulatively added. The motion vector for each detection block is calculated from the positional relationship between the sampling point and the representative point corresponding to the minimum value of the cumulative addition data calculated for each detection block. Four
By appropriately calculating the motion vectors for each block, the motion vector caused by the camera shake on the entire screen is obtained. According to this motion vector, the video extraction area in the imaging area is displaced, video information that follows camera shake is extracted, and the extracted video information is interpolated to form and derive a standard video signal. The configuration of the representative point matching described above is a well-known configuration that is widely used in a well-known video camera with a camera shake correction function. The representative point in FIG. 3 is set at a fixed position with respect to the image pickup area regardless of the occurrence of camera shake. Further, each detection block is displaced following the motion vector (difference vector) between fields, and the representative point is located at the center of the detection area before the occurrence of camera shake, but the representative point is located in the detection area after the occurrence of camera shake. It will be located off the center. The motion detection area is an area determined by the variable areas of all detection blocks. Coordinates in the figure (X ₀ ,
Y ₀ ) indicates the starting position of the first block in the initial state, the coordinates (X ₁₁ , Y ₁₁ ) are representatives corresponding to the first detection area of the first block, and the coordinates (X ₂₄ , Y ₂₄ ) are the first. Representative points and coordinates (X ₃₁ , Y corresponding to the 16th detection area of 2 blocks
₃₄₎ representative point corresponding to the 13 detection area of the third block, the coordinates (X _44, Y ₄₄₎ indicates a representative point corresponding to the 16 detection area of the fourth block, respectively.

In FIG. 3, the feature of the present embodiment is that the motion detection area is set in an area excluding the transient area of the vertical filter generated in the upper part of the imaging area and the transient area of the horizontal filter generated in the left side. It has a feature.

The specific operation of this embodiment will be described below with reference to the functional block diagram of the essential parts of the video camera of this embodiment shown in FIG.

First, the input signal obtained by image pickup is digitized by the AD conversion means in synchronization with the sampling clock. The digital image information is input to the horizontal filter 2 inside the camera shake removing means and band-limited in the horizontal direction. As a result, the video information is disturbed by the transient phenomenon at the beginning of the video period of each horizontal line. The transient region of the horizontal filter 2 in FIG. 1 is a region where the transient phenomenon of the horizontal filter 2 extends. The horizontal filter output is input to the vertical filter 3 and its output is stored in the line memory 4. The line memory 4 feeds back the stored output to the vertical filter 3 in a horizontal cycle, and the vertical filter 3 band-limits the filter output of the sampling point common in the vertical direction. As a result, the video information is disturbed by the transient phenomenon during the horizontal video period for a certain period following the vertical blanking. Figure 1
The vertical filter 3 transient region is a region where the transient phenomenon of the vertical filter 3 extends.

The output of the vertical filter is supplied to the representative point storage means 5 and the block-by-block correlation detection means 6. The representative point storage means 5 inputs only the filter output in the period in which no transient characteristic occurs, and stores the filter output level of the representative point. Similarly, the block-by-block correlation detection means 6 also inputs only the filter output in the period in which no transient characteristic occurs, and outputs the filter output of the detection block and the filter output of the representative point storage means 5 one field before the representative point. The absolute value of the difference between and is calculated, and the absolute value is cumulatively added in the same detection block to obtain the motion vector for each block.

FIG. 2 shows a concrete functional block diagram of the previous term representative point storage means 5. First, all the coordinates of the representative points are stored in advance in the representative point storage means 51. The start coordinate data of each detection area in the initial state is stored in the detection area start coordinate data storage means 53. The representative point storage means 51 supplies the sequentially stored representative point coordinate data to the first comparing means 52 and compares the horizontal count value with the X coordinate data or the vertical count value with the Y coordinate data. The horizontal count value is a value that is reset by the horizontal sync pulse and is derived by counting the sampling clock used for AD conversion. The previous vertical count value is reset by the vertical sync pulse and counts the horizontal sync pulse. The spatial position of the video information in the imaging area is represented by both values. Therefore, the storage timing pulse is generated from the first comparing means 52 in synchronism with the video information generating timing at each representative position, and is supplied to the write address generating means 55. The write address generating means 55 stores the vertical filter output for each sample in the representative point memory 57 when the storage timing pulse is generated. On the other hand, the detection area start coordinate data storage means 53 sequentially combines the X coordinate data and the Y coordinate data of the initial position data of each detection area and supplies them to the second comparison means 54. The second comparing means 54 also inputs the difference vector data V _NX V _NY in the X and Y directions, and each detection area of a field following coordinate data obtained by adding the motion correction data to the initial value data. It is the start coordinate data. The start coordinate data of each detection area is compared with the above-described vertical count value and horizontal count value in the second comparing means 54, and immediately after the start coordinate data is updated, the X coordinate match output and the Y coordinate match output are generated. It has occurred. Therefore, with respect to one Y coordinate coincidence output, the X coordinate coincidence output is generated eight times the number of lines in one detection area. Therefore, the read address generating means 56 interlocks with the generation of the Y coordinate coincidence output, and creates a combination of addresses in which eight representative points corresponding to eight spatially continuous detection areas are stored. Each time the X-coordinate coincidence output is specified, the address of the specified address set is sequentially updated and read continuously. As a result, from the representative point memory, the representative point data stored one field before corresponding to each detection area defined by the differential motion correction is sequentially derived as the correlation detection reference input.

Therefore, the block-by-block correlation detection means 6 compares the vertical filter output with the correlation detection reference input to obtain the absolute value of the difference, and determines the difference at the same position in each detection area in the common detection block. The absolute values are cumulatively added, the position on the detection area corresponding to the minimum cumulative addition value is specified for each detection block, and four types of block-based correction vectors connecting the center of the detection area and the specific position are obtained.

The block-by-block correction vector is input to the difference vector specifying means 7 to specify a correction vector for the entire imaging area by a predetermined calculation process, and is added to the previous difference vector to form and derive a new difference vector. .. This difference vector is a vector corresponding to the amount of camera shake between fields, and is used to specify the motion detection area in the next field as described above, and is also supplied to the camera shake correction data generation means 8. It The camera shake correction data generating means 8 cumulatively adds the difference vectors to form camera shake correction data, and supplies it to the read control means 9. The above-described horizontal filter 2 to the camera shake correction data generating means 8 correspond to the camera shake detecting means in the claims.

The read control means 9 includes a field memory 10
Only the video information of the video extraction area is read from the video information of all the imaging areas stored in advance, and the position of the video extraction area is changed by following the camera shake correction data. Therefore, the data read from the field memory 10 is the information of the video extraction area that follows camera shake,
The number of scanning lines and the length of the video period of each line are shorter than the original video information. The read control means 9 and the filled memory 10 correspond to the area selection means in the claims.

Therefore, the standard video data forming means 11 is
A small amount of read video information is interpolated and converted into standard video data. This standard video data is converted to analog by the DA conversion means 12 and is derived as a standard output video signal. As a result, the image formed by the output image signal becomes a stable image without camera shake.

[0020]

As described above, according to the present invention, when the camera shake is detected, the camera shake is detected in an area other than the transitional area of the vertical filter or the horizontal filter, so that the camera shake can be surely detected. , Its effect is great.

[Brief description of drawings]

FIG. 1 is a schematic functional block diagram showing an embodiment of the present invention.

FIG. 2 is a functional block diagram of essential parts showing a specific configuration of a representative point storage means of the present invention.

FIG. 3 is an explanatory diagram showing a relationship between an imaging area and a motion detection area.

[Explanation of symbols]

 2 horizontal filter 3 vertical filter 5 representative point storage means 9 read control means

Claims (1)

[Claims] 1. Filter means for band-limiting the spatial frequency of imaging information in the vertical direction or the horizontal direction, filter output levels of a plurality of representative points in one field or one frame before, and each representative point in the current field. A camera shake detection unit that calculates a correlation with a filter output level in the detection area and specifies a motion vector; and an image extraction area that follows the camera shake in the imaging area based on the output of the camera shake detection unit. In a video camera provided with area selection means for extracting video information from the video extraction area, a motion detection area constituted by each of the detection areas is set outside the filter transition area where the filter means exhibits a transient response characteristic. A camera shake correction circuit for a video camera characterized by being set to.