基于空间-时间多预测模式的无损压缩方法  

An Interframe Lossless Compression Method Based on the Mode of Space-Time Multi-Predictor

作　　者：周雨田[1] 左芝勇[1,2] 张天序[1] 兰霞[2] 姚守悝 邓丽华[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,武汉430074 [2]中国电子科技集团第十研究所,成都610036

出　　处：《计算机与数字工程》2015年第8期1481-1485,共5页Computer & Digital Engineering

基　　金：国家自然科学基金重点资助项目(编号:61433007);国家自然科学基金科学仪器专项资助项目(编号:61227007)资助

摘　　要：预测是无损压缩去相关的重要手段。针对卫星序列图像,考虑其在空间和时间上的相关性来改进JPEG-LS算法只是基于空间信息的预测方式,同时考虑到图像不同区域的不同特性,对图像进行分块处理。每个子块自适应选择一个最优的预测模板和预测方式,采用这个模板和预测器进行预测,使子块内的累计预测误差绝对值最小,从而获得较高的压缩比。Prediction is an important means of lossless compression decorrelation. For the sequence of satellite images, considering its relevance in space and time to improve the JPEG-LS algorithm which is only based on the spatial information, the image is divided into several blocks by taking into account different characteristics of different regions of the image. Each sub-block adaptively selects an optimal prediction template and prediction mode, in which the accumulated absolute value of prediction error is the smallest. The higher compression ratio is obtained by using the optimal template and predictor.

关 键 词：图像压缩 无损压缩 帧间压缩 JPEG-LS 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]