基于新特征和分形理论的快速分形图像压缩编码

发布时间：2018-04-04 11:09

本文选题：分形图像压缩　切入点：子块特征　出处：《南京邮电大学》2017年硕士论文

【摘要】：分形几何作为有力的数学新型工具,可以较好地完成对不规则图形的描画,使得其成为图像压缩编码方法的可选方向之一。然而,分形图像编码需要花费较长时间来搜索匹配块,且由于编码过程中会产生方块效应而影响到图像清晰度,所以,在保证图像质量不降低的前提下,提高图像编码速度成为本文需要讨论研究的课题。具体工作如下:首先,在特征向量法以及基本分形算法的基础上,提出一种新的子块特征表示方法(双交叉和特征)来定义规范块,再通过新特征对码本块进行排序,利用从赋序码本中找到的最佳匹配块进行匹配搜索。提出基于双交叉和的快速分形图像编码算法,由实验结果可得,该算法可在不影响图像质量的同时,可有效提高编码速度与相似度。其次,通过对于每个子块采用一种新的极差特征表示,并结合自适应搜索方法来搜索出最佳匹配块,给出了可行性分析,提出基于极差特征与自适应搜索相结合的分形图像压缩算法,通过对比实验证明,该算法在兼顾到重建图像质量和图像压缩编码耗时的问题上表现优良。最后,在叉迹特征算法以及最小均方误差与叉迹不等式的基础上,提出了一种间隔叉迹作为新特征的算法,分析其可行性,并进行仿真实验,实验结果表明,相对于一般叉迹特征,该算法对图像处理时间更少,而且重构图像质量也有保证。
[Abstract]:The mathematical model of fractal geometry as a powerful tool, can complete the irregular graphics drawing, making it become one of the directions of the image compression encoding method is optional. However, fractal image encoding takes a long time to search for a matching block, and the encoding process can produce the blocking effect and affect the clarity of the image. So, in the premise of guaranteeing the image quality is not reduced, improving the image encoding speed become the research topic to discuss in this paper. The specific work is as follows: firstly, based on the eigenvector method and basic fractal algorithm, proposes a new block feature representation method (double cross and features) to define the standard block, and through new features on this piece of code is sorted by find from the prefaces of codebook in the best matching block matching search. Proposed a fast fractal image coding algorithm and double crossover based on real Experimental results, this algorithm can not affect the image quality at the same time, can effectively improve the encoding speed and similarity. Secondly, for each sub block by using a new range of feature representation, and combined with the adaptive search method to search the best matching block, feasibility analysis is given, the fractal image compression algorithm the range and characteristics of adaptive based on the combination of the results of experiment show that the algorithm in both encoding and time-consuming problem to rebuild the image quality and image compression performance. Finally, the characteristics of the cross trace algorithm and based on minimum mean square error and cross trace inequality, proposes a new feature as cross trace interval the algorithm, analyzed its feasibility, and carries on the simulation experiment, the experimental results show that, compared with the general characteristics of the cross trace, algorithm of image processing in less time, but also ensure the reconstructed image quality.

【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.81;O189

【相似文献】