【作者】 陈洁；

【导师】 田逢春；

【作者基本信息】 重庆大学 ， 通信与信息系统， 2007， 硕士

【摘要】 21世纪,互联网技术正在迅猛的发展,以图像为主的多媒体技术大大丰富了我们的生活。但是如果没有一个高效的压缩方法,图像通信将不可能实现。图像压缩编码的目的就是要以尽量少的比特数表征图像,同时保持复原图像的质量,使它符合特定应用场合的要求。图像压缩也是多媒体技术的关键和瓶颈之一。分形图像处理技术是分形理论与图像处理技术结合的产物,已经在压缩编码、区域分割、模式识别等方面得到较多的应用。分形图像压缩方法是根据图像的自相似性,将一幅数字图像转化为一组收缩的迭代函数系统模型,通过对迭代函数系统参数编码达到图像压缩的目的。分形图像压缩方法具有压缩比高,解码速度快的优点。对于现实生活中的大量非严格自相似图像,常用的是基于子块划分的分形图像压缩方法。这种方法将图像划分为规则形状的不重叠的子块集,根据子块的局部自相似性,找出集中每一个子块的迭代函数系统,由全体子块的迭代函数系统参数形成分形图像压缩编码。传统分形压缩算法是在对子块(Range)和父块(Domain)进行了大小不同的分割后,在整个图像范围内寻找子块在压缩、仿射变换下的最佳匹配父块。由于每个父块一般都对应着8种仿射变换,所以搜索压缩映射块的过程消耗了大量的时间,压缩速度的缺陷足以抵消其优越性。本文以基于分形基本理论的图像编码研究为目的,在系统地介绍了分形压缩的基本理论后,针对传统方法耗时长的缺点,改进了邻域搜索的编码方法,即匹配块的搜索只在值域块的四邻域进行。这样大大减少了压缩的时间,同时也达到了压缩的目的。本文还基于人眼视觉特性的分形图像处理方法,分析了域块间相似性,减少了值域块搜索范围,在提高信噪比的同时,也大大减少了编码的时间。更多还原

【Abstract】 In the 21st century, the technology of network quickly develops, the multimedia technology based on image enriches our lives. However if there is no highly effective data compression approach, image communication can not be achieved. The purpose of image compression coding is to represent images with few bits, maintain the quality of recovering images according to the requirements of certain application situations. Image compression is the key and bottleneck of multimedia technology.Fractal image processing technique is widely applied in compression coding, region increasing, pattern recognition and so on. The method of fractal image compression transfers a digital image into a group of contract iterated function system (IFS) model. Encoding IFS’s parameters achieves image compression. This method may gain higher compression ratio, as well as rapid decoding. To many image which is not strictiy self-similar, the usual fractal image compression method based on block partition divides the image into non-overlap regular shape block collection. Every block’s iterate function system is found out by local self-similarity. The parameters of all iterated function system form fractal image compression code.Traditional fractal image coding first divides the image into ranges and domains at different size, and then searches the best matching domain of a range in the whole image after contracting and affine transforming of the domains. Since each domain commonly corresponds to eight affine transforms, the process of searching contracting mapping block cost vast time. The advantages of traditional fractal coding are counteracted by the low speed of compression. Therefore, after researching on basic theories of fractal compression, we improved algorithm based on nearest neighbor search. The new method search the domain block in the near neighbor. This procedure not only reduces the compression time, but also achieves the purpose of compression. In addition, we also introduce ideas based on human visual system about analyzing the similarity between blocks to improve the fractal image compression.更多还原