Current Issue Cover
邻域匹配和分类匹配的分形块编码

皮明红1, 彭嘉雄1(华中理工大学图象识别与人工智能研究所,武汉 430074)

摘 要
值域块的分形编码,是值域块与定义域块池中的定义域块逐一匹配后产生的。这一匹配过程是影响分形块编码速度的深层原因,此文改全局匹配为邻域匹配和分类匹配,从二条途径探讨了缩短匹配时间的方法。实验结果表明,领域匹配及分类匹配的分形块编码的时间,但信噪比略有下降。
关键词
Fractal Block Coding Based on Matching in the Neighbourhood and Classification of lmage Block

()

Abstract
Fractal code of a range block consists in find a best pair through matching between the range block and domain blocks in a pool of domain blocks one by one.This time consuming block matching procedure influences speed of fractal block coding.We focus on the reducing search time of block matching procedure by replacing full search matching with matching in the neighbourhood and matching based on classification of image block.Our result turns out that fractal block coding based on matching in the neighbourhood and classification of image block greatly reduces encoding time,but slightly degradation in PSNR of the decoded image.
Keywords

订阅号|日报