APPARATUS FOR DISCRIMINATING A MOVING REGION AND A STATIONARY REGION IN A VIDEO SIGNAL

Patent №

US 4,667,233

Granted

1987-05-19

Filed 1985

Owner

NEC CORPORATION

Lab

AI components

1

vision

Assignment

Recorded

Dataset

AIPD

2023_r1 edition

Application

06777033

An apparatus for discriminating a region that contains motion from a region that does not contain motion in a video signal, particularly for use in an inter-frame predictive coding apparatus. The apparatus bridges across different portions of a moving region or of a stationary region to avoid small isolated regions of either type that would cause spurious responses by the encoding circuit and a resulting degradation of picture quality. The apparatus determines for each picture element a frame difference, which is the brightness difference for that picture element from one frame to the next. The picture elements are divided into a predetermined plurality of blocks. Then a first evaluation value is produced by an adder (14, 14'), which may be alternatively the sum of the absolute values of the frame differences in each block, or the number of picture elements within each block for which the absolute value of the frame difference is greater than a predetermined threshold. In certain embodiments (FIGS. 4 and 5), a comparator (15, 15') determines whether the first evaluation value exceeds a predetermined threshold, and in response outputs a first moving/stationary region signal. The first moving/stationary region signal is stored in a memory (22) and processed by a decision circuit (23) which generates a final moving/stationary region signal which has been compensated to eliminate errors due to noise. The compensation by the decision circuit can follow either of two alternative methods, namely, spatial correlation between a block and the neighboring blocks; or time correlation between the status of a given block at particular points in time. The final moving/stationary region information signal employed to control an encoding circuit. In other embodiments (FIGS. 10 and 11) the first evaluation value is weighted by weighting factors to produce a second evaluation value. The weighting is based on the first evaluation values of blocks that are close to the given block either in time sequence or in location.

VisionH04N 19/503G06T 7/223G06T 2207/10016

AI classification

Vision0.77
Planning0.26
Machine learning0.06
Natural language0.03
Evolutionary computation0.00
AI hardware0.00
Knowledge representation0.00
Speech0.00

Ownership

NEC CORPORATION

assignment · 44980261

Assignors

FURUKAWA, AKIHIRO

On an employer assignment, the assignors are typically the inventors.

© 2026 NYSGPT2525 LLC