图像子块特征匹配的快速分形编码算法

发布时间：2018-05-27 16:35

本文选题：图像压缩 + 分形　；参考：《计算机工程与应用》2017年01期

【摘要】：基于分块迭代函数的全搜索分形图像编码算法,因其编码过程特别耗时而限制了它的诸多应用。为了减少编码时间,通过定义每个range块和domain块的子块特征,根据匹配均方根误差与它的关系,设计出一个限制搜索空间的新算法。一个待编码range块和它的最佳匹配domain块的子块特征应该接近,因此,每个range块的最佳匹配块搜索范围仅限定在与其子块特征接近的domain块邻域内,以达到加快编码过程的目标。14幅图像的仿真结果表明,该算法能够在PSNR降低0.73 d B(其结构相似性SSIM值仅下降0.002)的情况下,平均加快全搜索分形编码算法的编码速度99倍左右,而且也优于其他特征算法。
[Abstract]:The full search fractal image coding algorithm based on block iterative function limits its application because of its time consuming. In order to reduce the coding time, by defining the sub-block features of each range block and domain block, according to the relationship between the matching root mean square error and the matching RMS error, a new algorithm is designed to limit the search space. The subblock features of a range block to be coded and its best matching domain block should be similar, so the search range of the best matching block for each range block is limited to the neighborhood of the domain block that is close to its sub-block feature. In order to achieve the goal of speeding up the coding process, the simulation results of 14 images show that the algorithm can reduce the PSNR by 0.73 dB (the SSIM value of its structural similarity is only reduced by 0.002), and the coding speed of the full-search fractal coding algorithm is about 99 times faster than that of the full-search fractal coding algorithm. It is also superior to other feature algorithms.
【作者单位】：西南民族大学计算机科学与技术学院;
【基金】：四川省应用基础项目(No.2013JY0188) 四川省教育厅科研项目(No.15ZA0384)
【分类号】：TP391.41

【相似文献】