四分位数特征的快速分形图像编码算法

Fast fractal image encoding algorithm based on quartiles feature

尽管分形图像编码在高压缩比时具有较好的视觉质量,但其固有的编码耗时限制了它的广泛应用。编码时间主要花费于在一个海量码本中寻找每个range块的最佳匹配domain块。针对这个问题,提出一个限制搜索空间的算法:根据新定义的图像块的四分位数特征与匹配均方根误差间的关系不等式,一个待编码range块只在初始匹配块(即与range块具有最接近四分位数特征的domain块)的邻域内搜索其最佳匹配块,搜索邻域的大小由预先设置的误差阈值来控制。仿真结果表明,三幅测试图像在重建图像质量更优的情况下,全搜索分形编码算法的编码速度平均加快了51倍左右(误差阈值为10),与新叉迹特征算法相比,可获得更好的编码效果。

Although fractal image coding has excellent visual quality at high compression ratio,it has not been widely applied due to exhaustive inherent encoding time.The time is mostly spent on searching for the best-matched block to every range in a usually large domain pool;therefore,a fast algorithm is proposed to limit the search space in this paper.The search scope of best-matched block for an input range block is local against full on the basis of an inequality linking the root-mean-square and newly-defined quartiles feature of normalized block.In detail,it can effectively confine the search space to the vicinity of the initial-matched block（i.e.,the domain block having the closest quartiles feature to the input range block being encoded）.Besides,an error threshold is also used to control the size of search neighbourhood automatically.Simulation results demonstrate that,for three standard test images,the proposed scheme averagely obtain the speedup of 51 times or so by error threshold set 10,while can accomplish good quality of the reconstructed images against the full search method.Moreover,its performance is better than the newly cross trace feature of normalized block algorithm.