CAJ | 학술논문

针对感兴趣区域优先编码的问题,提出了基于粒子滤波目标跟踪的精细可分级视频编码(FGS,Fine Granularity Scalable),该方法结合开源计算机视觉库(OpenCV),在增强层引入粒子滤波运动跟踪算法,人为定义出一个感兴趣的区域,对此区域进行跟踪,提取,位平面提升,优先传输和编码,并控制优先编码的位平面数,控制预测漂移,从而改善视频解码后的主观视觉效果.同时,在基本层引入单环算法来提高编码效率.实验结果表明,提出的算法在一定的码率范围内感兴趣区域(ROI)的PSNR值高于整体区域的PSNR值.
침대감흥취구역우선편마적문제,제출료기우입자려파목표근종적정세가분급시빈편마(FGS,Fine Granularity Scalable),해방법결합개원계산궤시각고(OpenCV),재증강층인입입자려파운동근종산법,인위정의출일개감흥취적구역,대차구역진행근종,제취,위평면제승,우선전수화편마,병공제우선편마적위평면수,공제예측표이,종이개선시빈해마후적주관시각효과.동시,재기본층인입단배산법래제고편마효솔.실험결과표명,제출적산법재일정적마솔범위내감흥취구역(ROI)적PSNR치고우정체구역적PSNR치.
@@@@For the preferential encoding region of interest, the FGS(Fine Granularity Scalable)coding strategy based on particle filter video tracking is improved. This strategy combines Open CV (Open Source Computer Vision Library) and introduces the particle filter video tracking algorithm into the enhanced layer. An area artificially defined as an area of interest is tracked and extracted, and bit-plane upgrade, preferential transmission and coding are also done, including the control of bit-plane number and prediction drift, thus to improve subjective visual effect after video decoding. Meanwhile a single-loop algorithm is introduced into the basic layer for improving the objective coding efficiency. The experiment shows that the PSNR value of ROI(region of interest)is higher than that of overall regional within certain bit-rate range.