基于小波变换的多方向分形预测图象编码方法

小波图象编码属于子带编码，但是，与通常的子带编码不同，图象经过小波变换后，不但分解成不同频带（即分辨率）的子带图象，而且，同分辨率的子带图象还进一步分解成不同空间方向的子带图象。同方向的各个子带图象之间具有相似性，利用这种相似性进行图象编码可以提高压缩比。不同方向的各个子带图象之间虽然不具有相似性，但是，原图象的特征在这些子带图象中的空间位置具有不变性，利用这种空间位置的不变性进行图象编码，可以进一步提高压缩比。本文提供了这方面的理论分析和实例验证。

Wavelet-Based Multi-Orientation Fractal Predictive Image Coding

Wavelet image coding is a kind of sub-band image coding. However, differing from commonly used sub-band image coding methods, an image, when decomposed with wavelet transform, is divided into a number of sub-images not only of different frequency bands (resolutions), but of different spatial orientations. In order to gain higher compression ratios, The neighboring sub-images of the same orientation are strongly alike. The sub-images of different orientations, though few similarities are presented between them, use the same spatial area to represent the characteristics of the same object of the original image. Therefore, this paper presents an approach which exploits the invariance of spatial locations thus obtains an additional increase in compression ratio. Experiment results also show that using this proposed method, reconstructed images with higher quality and better compress ratios can be obtained.