基于串匹配的多视点视频图像阵列自编码仿真  

Self-encoding Simulation of Multi-View Video Image Array Based on String Matching

作　　者：赵全宜[1] 张泽 ZHAO Quan-yi;ZHANG Ze(Hubei University of Technology,Hubei Wuhan 430068,China;School Of Industrial Design Hubei University Of Technology,Hubei Wuhan 430068,China)

机构地区：[1]湖北工业大学,湖北武汉430068 [2]湖北工业大学工业设计学院,湖北武汉430068

出　　处：《计算机仿真》2021年第12期146-149,181,共5页Computer Simulation

基　　金：2017年度湖北省教育厅人文社会科学研究项目青年项目(17Q073)。

摘　　要：针对传统多视点视频图像列阵自编码方法编码效率低、响应时间长的问题,提出基于串匹配的多视点视频图像阵列自编码方法。首先根据串匹配算法,组建视点之间相互对应的对极线校正索引表。将其应用于视觉估计中来缩小两视点之间的视差搜索范围,然后将原有的视差搜索二维降到一维,利用拟合三维二次函数确定不同关键点的具体坐标位置以及尺度,同时删除无用的响应点,在得到匹配点后,通过描述子来准确描述特征点,匹配不同的描述子信息,获取符合标准的匹配点集,以达到多视点视频图像阵列自编码的目的。仿真结果表明,所提方法能够有效提升编码效率,且所需响应时间短,能够获取理想的编码效果。Aiming at the problems of low coding efficiency and long response time of traditional multi view videoimage array self coding method, a multi view video image array self coding method based on string matching is pro-posed. According to the string matching algorithm, the epipolar line correction index table corresponding to eachviewpoint was constructed. Then, it was applied to visual estimation to narrow the disparity range between two view-points. After that, the original 2 D disparity search was reduced to one dimension. Moreover, the specific coordinatepositions and scales of different key points were determined by fitting the 3 D quadratic function. Meanwhile, the use-less response points were deleted. After the matching points were obtained, the feature points were accurately de-scribed by descriptors. Finally, different descriptors were matched to get the set of matching points that met the stand-ard, and thus to achieve the self-encoding of multi-view video image array. Simulation results prove that the pro-posed method can effectively improve the coding efficiency, with lee response time. Meanwhile, the coding effect is better.

关 键 词：串匹配 多视点视频图像阵列 自编码 对极线校正索 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]