提出了一种覆盖式分形压缩方法,并给出了相应快速搜索算法.该方法采用四连通扩充的方式,在给定误差范围内,尽可能地合并规则Range分区,从而减少所需迭代变换的数目.为减少存储Range分区信息所需的字节,在存储压缩映射时,根据Range分区多边形的边数调整存储顺序,对于那些复杂的Range分区,只记录其外接矩形框信息.在解码时,采用一种覆盖式绘制方法,可以完全恢复原Range分区情形.通过对Lenna,图象的实验表明,在恢复图象的信噪比为29.63dB时,可以达到41倍的压缩倍数.

The paper presents a new fractal image compression scheme,which merges the range blocks as much as possible in a 4-connected extending way to decrease the amount oI the transformations under a given error tolerance,with a quick searching algorithm. Through rearranging the storing order of the transformations according to their edges,we only need to record their surrounding rectangles for those complicated ranges blocks .whose shape can be recovered using a paint-like rendering method in decoding.Our experiment on Lenna is pleasing.the compression ratio reach 41 at a PNSR of 29.63dB.