Fractal Image Compression Based On Range Blocks Classification

People can store and transmit an enormous amount of information by making rational and efficient use of image .During this process ,image compression is an important step .The idea of fractal image encoding originates from Barnsley's paper .Since then ,a number of papers on fractal image encoding have been published .Among this paper ,Jacquin's is the most important one. By his scheme ,the process of fractal image compression has been fully automated.Jacquin's basic fractal compression scheme partitions an image into two kinds of blocks .One is range blocks ,the other is domain blocks. We need encode each range block by finding the best affine transformation for it.Fractal image encoding has the advantage of very fast decompression and potentially very high compression ratios .These properties made it a very attractive method for applications in multimedia .But it's encoding time is very long . This is the important reason which prevents the method from being well received by people.We all know the encoding time is very long .the reason is the number of the global domain block pool is very large . In order to decrease the encoding time , we need to reduce the number of domain block that must be compared with the range block .One of these method is searching the nearest neighbor of the range block.In this paper ,we will propose an search scheme. First we will do a pre-searching within the range block pool .We will find the central block and the similar blocks whose characteristic is similar with the range block .Then we will find the best affine transformation for the central block .But the searched area will not be the global domain block pool .The area that need to be searched should be located in the nearest area of the similar blocks;Experimental results showed this technique is able to reduce encoding time if we had chose a proper threshold.