Television – Image signal processing circuitry specific to television – Motion vector generation
Patent
1993-10-08
1996-09-17
Chin, Tommy P.
Television
Image signal processing circuitry specific to television
Motion vector generation
348415, 348411, 348412, H04N 514, H04N 964
Patent
active
055573410
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
The invention relates generally to a method for estimating motion contents in video signals, said motion being normally represented by a two-dimensional motion vector field, including horizontal and vertical vector components. The general purpose of such an estimation is to allow high quality temporal transformation or processing of video signals. The invention can be applied in a variety of fields, some important examples being: conversion between different television signal standards, noise reduction of video signals, data compression for digital transmission of television or high definition television signals, and slow-motion devices.
Known methods of generating motion vector fields include recursive pixel gradient methods, phase correlation and block matching techniques. Each technique has its own set of advantages and disadvantages with respect to the following generally desirable features: high speed, as a result of using a large search area; description of the motion vectors of moving objects, as a result of discriminating different velocities and directions of movement of neighbouring blocks. movement of small adjacent areas of moving objects, as a result of discriminating neighbouring pixels with different movement. an object by generating a motion vector field while suppressing noise and other disturbances; and implementation) involving the ability to use the method commercially for a wide range of applications.
The advantages and disadvantages of the known methods appear from the table below:
______________________________________ Standard
Pixel Phase block
gradient
correlation
matching
______________________________________
A: High tracking range
NO YES YES
B: High vector resolution
YES YES YES
C: High spatial resolution
YES NO NO
D: Robust algorithm
NO YES NO
E: Cost effectiveness
YES NO YES
______________________________________
In the standard block matching technique (compare, e.g. U.S. Pat. No. 4,924,310), the video signal is represented by a consecutive sequence of pictures separated in time, and said pictures are divided into blocks of predetermined orientation and size. The block size will determine the spatial resolution of the vector field. To enable comparison or matching of blocks, one has to define a search area with a certain size and shape, a search pattern, according to which the search or comparison will be carried out, and a matching criterion representing the degree of likeness of one block to another. Such a criterion is based on a comparison of light intensity or luminance of the individual picture elements (pixels).
For the purpose of determining the momentary motion vector field, two consecutive pictures are regarded. The blocks, made up of a number of pixels of the current consecutive picture, are processed in order. For every such block a search is carried out in the search area of the previous consecutive picture, according to the search pattern, so as to match said block to a block in the previous consecutive picture. The direction and the value of the relative displacement of the two matching blocks are selected as the motion vector of the block in the current picture. A complete momentary motion vector field is obtained after blockwise processing of the entire content of the current picture.
SUMMARY OF THE INVENTION
The present invention concerns an improved method using the block matching technique, and is specially adapted for cost effective real-time applications.
The particular problem to be solved by this improved method of block matching is to increase the spatial resolution and to provide robustness in the algorithm. According to the invention, the algorithm is based on hierarchical processing with different block sizes from coarser to finer with the calculation of predicted motion vectors from surrounding blocks and, preferably, post processing by non-linear filtering of the motion vector field after each hierarchical search step, as well as motion adaptive temporal filtering after calculation of the motion
REFERENCES:
patent: 4691230 (1987-09-01), Kaneko et al.
patent: 4796087 (1989-01-01), Guichard et al.
patent: 4800425 (1989-01-01), Schwerzel et al.
patent: 4853779 (1989-08-01), Hammer et al.
patent: 4924310 (1990-05-01), von Brandt
patent: 4937666 (1990-06-01), Yang
patent: 4989087 (1991-01-01), Pele et al.
patent: 5151784 (1992-02-01), Lavagetto et al.
patent: 5173772 (1992-12-01), Choi
patent: 5193001 (1993-03-01), Kerdranvrat
"A Spatio-Temporal Gradient Method for Estimating the Displacement Field in Time-Varying Imagery," Paquin, et al., Computer Vision, Graphics, And Image Processing, No. 21, pp. 205-221 (1983).
"Displacement Measurement and Its Application in Interframe Image Coding," Jaswant R. Jain, IEEE Transactions on Communications, vol. COM-29, No. 12, pp. 1799-1808, Dec. 1981.
"Motion-Compensated Television Coding: Part I," Netravali et al., The Bell System Technical Journal, vol. 58, No. 3, pp. 631-670, Mar. 1979.
Christensson Bjorn
Weiss Peter
Chin Tommy P.
DV Sweden AB
Rao Anand S.
LandOfFree
Iterative method for estimating motion content in video signals does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Iterative method for estimating motion content in video signals , we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Iterative method for estimating motion content in video signals will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-416719