摘要
描述的子空间投影技术(SPT)是一种把视差补偿(DC)和二维低阶近似相结合的立体图像编码方法。这种算法可以随着立体对交叉相关特性的改变而局部自适应。在视差估计失败的遮挡区域,由低阶近似提供比较理想的投影图像,并通过零阶项对立体对的光学度变化进行补偿。实验结果表明,SPT方法其峰值信噪比(PSNR)高于像块尺寸固定的视差补偿算法。
This paper describes a stereo image coding algorithm---the subspace projection technique(SPT),which is based on disparity compensation and two-dimensional low order approximation.The advantage of the pro-posed approach is that it can locally adapt to the changes in the cross-correlation characteristics of the stereo pairs.For the regions where disparity estimation fails,such as occlusion regions,low order approximation provides a better projecting images.And zeroth order term compensates for photometric variations.Simulation results shows that the SPT approach outperforms the fixed block size schemes in PSNR gain.[
出处
《电视技术》
北大核心
2002年第7期24-26,共3页
Video Engineering
基金
国家自然科学基金资助项目(69972027)
关键词
立体图像
编码
子空间投影
视差估计
视差补偿
stereo image coding
disparity estimation(DE)
disparity compensation(DC)
SPT
