摘要
针对感兴趣区域问题,对增强层的编码加以改进,提出基于运动对象跟踪的FGS(Fine Granularity Scala-ble)视频编码方法.该方法利用了OpenCV(开源计算机视觉库),实现运动对象的检测与提取,通过提取,将运动对象作为感兴趣区域,进行位平面提升,优先传输.并在基本层上引入提高编码效率的单环算法,进一步提高视频主观与客观质量.实验结果表明,算法在大部分的码率范围内ROI区域的PSNR值高于整体区域的PSNR值,对背景静止的视频序列,效果理想.
Targeted to the problem of region of interest,we improve the enhanced layer coding strategy and propose the FGS（Fine Granularity Scalable） video coding based on moving object tracking.This method make use of OpenCV（Open Source Computer Vision Library）,achieving moving object detection and extraction.By extracting we make the moving object as ROI,enhancing the Bit-plane,transmission first.And introducing a single-loop algorithm which on the base layer,further enhancing the subjective and objective quality video.The results of experimental show that the algorithm in the context of ROI region PSNR value higher than the overall PSNR value in Most of the bit rate,the background static video sequences,the results are satisfactory.
出处
《微电子学与计算机》
CSCD
北大核心
2010年第12期134-137,共4页
Microelectronics & Computer
基金
国务院侨办科研基金项目(06QZR03)
