基于稠密矢量场及自适应补偿的帧率上变换  

Frame Rate Up-conversion Based on Dense Vector Field and Adaptive Compensation

作　　者：刘梦可 周军[1] 高志勇[1] 陈立[1] 

机构地区：[1]上海交通大学图像通信与网络工程研究所,上海200240

出　　处：《电视技术》2015年第11期11-14,93,共5页Video Engineering

基　　金：国家科技重大专项项目(2013ZX01033001-002-002);国家自然科学基金项目(61133009);数字媒体处理与传输上海市重点实验室项目(STCSM 12DZ2272600)

摘　　要：为了有效解决视频帧率上变换中物体边缘出现的模糊和块效应等问题,结合运动估计,优化了矢量化后处理和运动补偿算法。在基于块的运动估计中,结合运动的时空相关性优化块匹配算法,提高物体边缘运动矢量的准确性;通过基于像素块相关性的插值算法,获取像素矢量场,并保持物体结构完整性;在运动补偿时,结合矢量后处理的分类信息,自适应地进行加权插值计算,提高内插帧的质量。实验结果表明,与传统方法相比,该方法在主观和客观质量上都有较大的提升。In order to solve problems such as edge blur and blocking artifacts in frame rate up-conversion,block matching （BM） algorithm and motion compensation （ME） algorithm are improved based on motion estimation.In the block-based motion estimation way,BM algorithm is optimized by adding motion information of temporal and spatial correlation,and reducing the blocking effect of moving objects.Pixel MV is applied in the MC and is calculated by bilateral filter based on correlation between pixel blocks,to maintain the structural integrity of the objects.In motion compensation,the classification information of the MV processing is used for adaptively weighted interpolation to improve the quality of the interpolated frame.Experimental results show that,the image quality is improved in the subjective and objective quality.

关 键 词：帧率上变换 块匹配 双边滤波 自适应加权插值 

分 类 号：TN911.73[电子电信—通信与信息系统]