Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1998-05-27
2003-02-04
Lee, Richard (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
Reexamination Certificate
active
06516031
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a motion vector detecting device favorably used in an image coding apparatus according to MPEG (Moving Picture Experts Group), and more particularly to an improvement for reducing a hardware volume.
2. Description of the Background Art
To transmit or store signals representing an image, i.e., image signals, having a large amount of data, an image compression technique for reducing the amount of data is essential. The image signal, in general, includes considerable redundancy resulting from correlation between adjacent pixels and human perceptional characteristics. The image compression technique for suppressing the redundancy to reduce the amount of data is termed high-efficiency coding. In the present specification, to avoid useless complicate description, an image signal representing an image and an image itself are equally represented as “image” if not leading to confusion.
As one of such high-efficiency coding systems known is an inter-frame predictive coding system. In this inter-frame predictive coding system, a predictive error, which is a difference between a current image (frame or field) to be coded and a reference image (frame or field) to be referred to, being the preceding or following data in terms of time, is calculated for each pixel of the same position. The predictive error as calculated is thereafter used for coding.
In this system, it is advantageously possible to perform a coding of an image having small movements at high efficiency, due to high correlation between frames (fields). As to an image having large movements, however, errors are disadvantageously increased due to small correlation between frames (fields), leading to an increase in amount of data to be transmitted or stored.
Further, as a system to solve the aforementioned problem known is a motion compensated inter-frame (or inter-field) predictive coding system. In this system, prior to calculation of the predictive error, a motion vector is calculated by comparison of pixel values between the current image and the reference image. The motion vector refers to a vector representing magnitude and direction of an inter-frame (or inter-field) local movement of an image.
A local image in the reference image is moved in accordance with the motion vector as calculated, and then used as a predictive image for the calculation of the predictive error. Specifically, the predictive error is calculated by comparing the predictive image obtained by motion-compensating the reference image with the current image. Then, the predictive error as calculated is used for the coding. The motion vector along with a coding signal corresponding to the predictive error is transmitted or stored as an image signal after compression.
Thus, the motion compensated inter-frame (or inter-field) predictive coding system has an advantage of achieving a high compression efficiency of an image, even if having a large movement, since the inter-frame (or inter-field) predictive coding is performed by adopting the motion compensation using a motion vector.
The detection of the motion vector (predictive image) in the inter-frame (or inter-field) predictive coding system, however, requires enormous amounts of operations, as shown in the document (Mitsubishi Denki Gihou: Vol. 69, No. 3, pp. 12 to 16 (1995), “A Motion-Estimation LSI with Half-Pixel Precision”) and as a result, there is a problem of requiring enormous amounts of hardware to implement a wide-range search of motion vectors.
SUMMARY OF THE INVENTION
The present invention is directed to a motion vector detecting device for searching a predictive block included in a reference image, for which a predictive error should be obtained by comparison with a template block included in a current image, and detecting a motion vector indicating a displacement between the template block and the predictive block. According to a first aspect of the present invention, the motion vector detecting device comprises: an operation unit for calculating estimation values each indicating an overall similarity of pixel values between the template block and a displacement block every time when the displacement block is moved in a search window after skipping pixels of the template block and the displacement block, the displacement block having the same size as the template block and included in the search window, and the search window being defined in the reference image as a region including a block corresponding to the template block; a comparison unit for searching a minimum value among the estimation values on the template block and outputting the motion vector corresponding to a displacement block providing the minimum value, which is defined as the predictive block; and an input unit for supplying the pixel values of the template block and the pixel values of the search window to the operation unit.
According to a second aspect of the present invention, in the motion vector detecting device of the first aspect, the input unit comprises a filter unit for performing a filtering of the current image and the reference image and supplying the pixel values of the template block and the pixel values of the search window to the operation unit.
According to a third aspect of the present invention, in the motion vector detecting device of the first aspect, the input unit comprises a filter unit for performing a filtering of the current image and the reference image in an on/off selectable manner in response to an externally-received control signal and supplying the pixel values of the template block and the pixel values of the search window to the operation unit.
According to a fourth aspect of the present invention, in the motion vector detecting device of the second aspect, the filter unit performs the filtering by converting a pixel value of an object pixel to be filtered into a weighted mean of pixel values of a plurality of consecutive pixels along a horizontal scanning-line direction including the object pixel, with an weight to be added to pixels outside a block including the object pixel set zero.
According to a fifth aspect of the present invention, in the motion vector detecting device of the first aspect, the operation unit comprises a template-block storing unit for storing pixel values of all pixels included in the template block after skipping; a search-window storing unit for storing pixel values of all pixels included in the displacement block; an estimation-component-value operation unit for calculating estimation component values for all the pixels included in the template block after skipping, each of the estimation component values indicating a similarity between the pixel values stored in the template-block storing unit and the pixel values stored in the search-window storing unit on a pixel-by-pixel basis; a summation unit for synthesizing the estimation component values to calculate the estimation values, the search-window storing unit comprises search-window storing elements arranged in a matrix, the search-window storing elements being so cascaded and each storing a pixel value for one pixel as to move the displacement block in the search window by storing the pixel values of the search window in order, and the input unit supplies the pixel values of the template block to the template-block storing unit and supplies the pixel values of the search window to the search-window storing unit.
According to a sixth aspect of the present invention, in the motion vector detecting device of the fifth aspect, the template-block storing unit comprises a (n, m)-th template-block storing element for storing a pixel value for one pixel, where n=1 to N (N is an even number) and m=1 to M (M is an even number), the search-window storing unit comprises a (k, m)-th search-window storing element for storing a pixel value for one pixel, where k=1 to 2N and n=1 to M, in a plurality of columns of search-window storing elements and (1, m)-th to (2N, m)-th search window elements are cascaded i
Hanami Atsuo
Ishihara Kazuya
Kumaki Satoshi
Matsumura Tetsuya
Matsuura Yoshinori
Lee Richard
Mitsubishi Denki & Kabushiki Kaisha
LandOfFree
Motion vector detecting device does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Motion vector detecting device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Motion vector detecting device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3137028