CAJ | 학술논문

传统的分形图像编码根据图像内部的跨尺度自相似性，求解压缩仿射变换，完成图像编码，这种来自于自身的编码码本不一定是最佳的，而且编码时间通常较长。根据图像之间存在着相似性，构造一个适应性更强的固定编码码本，其它任何图像分形编码的值域块只在这个固定码本中搜索其最佳匹配块，并且省去等距变换，此外，解码也无需迭代。实验结果表明，在图像质量略有下降情况下，该算法较基本分形图像编码算法显著地提高了编码和解码的速度，而且算法简单，容易实现。如果对这种固定码本加以更好的适应性改造，将会进一步提高解码图像的质量。
전통적분형도상편마근거도상내부적과척도자상사성，구해압축방사변환，완성도상편마，저충래자우자신적편마마본불일정시최가적，이차편마시간통상교장。근거도상지간존재착상사성，구조일개괄응성경강적고정편마마본，기타임하도상분형편마적치역괴지재저개고정마본중수색기최가필배괴，병차성거등거변환，차외，해마야무수질대。실험결과표명，재도상질량략유하강정황하，해산법교기본분형도상편마산법현저지제고료편마화해마적속도，이차산법간단，용역실현。여과대저충고정마본가이경호적괄응성개조，장회진일보제고해마도상적질량。
The traditional fractal image coding is based on the self-similarity across different scales of the image inside, solves the compression affine transformation and completes the image coding. The codebook from the image itself is not necessarily the best, and the encoding time is usually longer. According to the existence of similarities between the images, we construct an adaptable codebook in which the best matching block can only be searched for by the range block of any image fractal coding, and isometric transformation is omitted. In addition, there is no need to iterate the decoded image. Experimental results show that in the case of a slight decrease in image quality, the algorithm has better performance than basic fractal image coding and improves the encoding and decoding speed, and the algorithm is simple and easy to implement. If the fixed codebook improves through a better adaptation, it will enhance the quality of the decoded image.