基于空间-时间多预测模式的无损压缩方法

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 .