基于图像边缘度的零树压缩最优分解层的选择  

The Optimal Selection of Decomposition Level in EZW Based on Edge Active Measure

作　　者：张旗[1] 梁德群[2] 

机构地区：[1]大连海事大学自动化与电气工程学院,大连116026 [2]大连海事大学信息工程学院,大连116026

出　　处：《中国图象图形学报》2005年第12期1491-1497,共7页Journal of Image and Graphics

摘　　要：在研究零树图像编码方案的基础上,分析了被压缩图像的空间冗余度(称为图像边缘度)与零树压缩结果的关系,提出了零树压缩存在小波最优分解层的概念,给出了不同压缩比、不同小波分解层与不同类型图像的零树压缩结果的关系,并对相应的压缩时间进行了分析。综合实验结果表明:在进行高比特率压缩时,提倡采用三层小波分解;而进行低比特率压缩时,应采用四层小波分解。这一结论可作为进一步研究具有自适应能力的改进零树算法的基础。Based on the investigation of EZW, the relationship between space redundancy （ scaled by edge active measure ） of the compressed image and the result of EZW coding is analyzed, and the concept that there exists an optimal wavelet decomposition level in EZW is proposed. The relationship in EZN coding between different compression ratio, different decomposition level and different images is reported. The executing time is also used as a factor for selecting decomposition level. The experimental results indicate that the three-level decomposition in EZW should be adopted for high compresssion ratio and the four-level decomposition is optimal for low ratio. The results provide a basis for further research on adaptive EZW algorithms.

关 键 词：图像边缘度 零树压缩 小波分解层 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TN919.81[自动化与计算机技术—计算机科学与技术]