Dynamic zooming search window-based motion estimation method and system

Abstract

The invention discloses a dynamic zooming search window-based motion estimation method and a dynamic zooming search window-based motion estimation system. The method comprises the following steps of: performing zero motion pre-termination judgment on a video sequence; predicting initial search points; performing initial positioning by searching on a large scale by adopting an extended hexagon search method; performing secondary search in a determined range on a small scale by adopting a diamond search method; and further performing precise positioning in the determined range by adopting a cross search pattern (CSP) method to obtain a motion vector. The system is characterized by consisting of a video data input module and a central processing unit module, wherein the video data input module is formed by connecting an acquisition charge coupled device (CCD) camera with an analogue-to-digital (A / D) converter in series; the central processing unit module comprises an image preprocessing module, a first buffer, a second buffer, a search window data memory, a comparator, a control subprogram control module, an operation subprogram control module and a clock control module. The method and the system improve the search efficiency of a motion estimation algorithm.

Description

technical field [0001] The invention relates to a video image compression coding motion estimation method, which belongs to the field of video image processing. Background technique [0002] Today, when information interaction is highly developed, video interaction and distribution services using the network as the carrier, such as network video conferencing, network visual e-commerce, and online government affairs, have put forward higher requirements for the expression, storage, and transmission of video information. Almost all modern Video processing technology needs to face the problem of massive data. How to deal with massive data, reduce the complexity of video expression, reduce the hardware cost of video storage, and improve the real-time performance of video transmission are the research focus of today's video processing technology. In order to overcome these difficulties, the traditional processing strategies mostly use methods such as expanding memory capacity an...

AI Technical Summary

Problems solved by technology

It is precisely because of this large-scale change in the position of the macroblock between frames that the search strategy of the traditional motion estimation algorithm is difficult to quickly determine the vector range and determine the vector end point with high precision. When estimating the motion vector of the target in a high-speed motion scene, Either the complexity of the block matching search operation increases sharply, or the search accuracy decreases significantly, which eventually leads to a decrease in video quality after compression

Images