利用图像自相似性进行编码的研究

遥感图像的平坦区域中存在大量冗余信息,探索更有效去除这些区域内冗余信息的方法对于提高图像编码效率有很大帮助。通过引入图像自相似性的概念进行研究,发现在图像的平坦区域中存在大量的自相似部分。从图像自相似性的角度出发,利用图像的局域自相似性对平坦区域进行编码以提高编码效率。在基于模式特征的遥感图像压缩算法中利用这种方法,针对一组典型遥感图像进行压缩,使高倍率压缩下恢复图像的PSNR平均提高0.1 dB左右。从不同图像的压缩结果来看,增加自相似的方法对于去除较平坦区域中的冗余信息相当有效。

Research on the image coding by using self-similarity of image

There are large amounts of redundant information in the smooth areas of remote sensing images.To find a more effective way to reduce such information is very helpful to improve the result of image compression.With image self-similarity being used for studying,it can be found that lots of small flat areas are much similar to each other.Since the self-similarity is the inherent character of an image,using the local similarity of an image is a more effective way for compressed coding of the small flat areas.The idea mentioned above is proved to be effective by incorporating the self-similarity coding into the pattern character based data compression algorithm.The compression results of several typical remote sensing images show that the PSNR(power signal-to-noise ratio)value of reconstructed images can be increased by 0.1 dB or so.From the compression results of different images,it can be concluded that using self-similarity of image can reduce the redundant information in the smooth areas of image effectively.