基于局域纹理特征的图像无损压缩

局域纹理是识别图像的基本要素,因此在图像压缩处理中分析局域纹理特征是达到去相关的有效途径,在此原理基础上提出了一种新的基于局域纹理特征的无损压缩方法,该算法由去相关处理和熵编码两部分组成,去相关操作是通过一组预测模型自适应选择最佳的模型进行处理,然后使用快速有效的CCSDS编码器对最佳去相关模型的预测误差进行编码,最后给出了该算法与JPEG2000及JPEG-LS的无损压缩结果的比较,实验结果显示,该算法的压缩结果明显优于JPEG2000及JPEG-LS的无损压缩结果. 

Lossless Image Compression Based on Feature of Local Texture

Local texture was the basic element of recognizing the whole image and starting with analysis of local texture was a reasonable way in developing better decorrelation model. Based on this idea, a new lossless image data compression algorithm named lossless image compression based on feature of local texture (FOLT) was presented. The algorithm was composed from two parts, decorrelation and entropy coding. Decorrelation was based on adaptively selection best decorrelating model suitable for the processed local area from a given model group. And the CCSDS adaptive entropy coder was used for entropy coding. The compression results not only show that the performance of the proposed algorithm is better than that of JPEG LS,but also show the proposed decorrelation idea is effective.