Font Size: a A A

Research And Implementation Of Video Object Segmentation In Videoconference System

Posted on:2005-08-10Degree:MasterType:Thesis
Country:ChinaCandidate:Z WangFull Text:PDF
GTID:2168360122996789Subject:Signal and Information Processing
Abstract/Summary:
The research of video coding is mainly focused on two aspects in recent years, one is the very low-bit rate coding, the other is the video coding based on objects. People are concerned about how to improve the objective quality of coding images for the former. The latter offers higher interactivity for multimedia services such as videoconference and net shopping, it is the current research focus and is greatly impelled by the establishment of MPEG-4 standard. Based on the existing videoconference system of Ministry of Communications (MOC),this paper farther discusses the correlative video technology of it.The videoconference system sponsored by MOC adopts the H. 263 video coding standard and it can achieve the bit-rate between 64kb/s and 1.92Mb/s. MPEG-4 is a standard aimed at the very low bit-rate motion pictures and speech, and object is the basic unit of MPEG-4 standard. If we can implement the segmentation of video object, the very low bit-rate video coding and camera tracking are then expected. This paper mainly discusses the video object segmentation technology. In addition, the functions of some sub-modules in videoconference system are also improved.This paper is divided into five parts. The first part introduces the background of this paper and my tasks. The current segmentation algorithms aiming at video coding are included in the second part. In the third part, I implement the segmentation of video object in videoconference sequences and present a method for motion segmentation based on orientation information measure. The fourth part mainly discusses two sub-modules of videoconference system, in addition, the test and evaluation of system are also included. The last part is the conclusion and expectation.
Keywords/Search Tags:videoconference system, MPEG-4, video object segmentation
Related items