多义图像合成算法研究  

Algorithm for Multi-Meanings Image Synthesis

作　　者：刘文兆[1,2] 湛永松[3] 石民勇[2] 

机构地区：[1]天津师范大学新闻传播学院,天津300387 [2]中国传媒大学数字技术与艺术研发中心,北京100024 [3]桂林电子科技大学计算机与控制学院,广西桂林541004

出　　处：《工程图学学报》2011年第4期31-37,共7页Journal of Engineering Graphics

基　　金：天津师范大学基金资助项目(5RW087);国家广电总局科技资助项目(2007-07)

摘　　要：提出了一种基于小波变换的多义图像合成算法。该算法首先通过视角来选择小波分解级对两幅目标图像进行小波分解,然后根据可选的小波融合规则和权值变量融合并重构出在两个视角下的多义图像;对小波算法与现有两种算法进行了分析比较,通过试验证明该算法合成的结果多义特性更好,合成图视觉效果醒目并充分保留原目标图像的色彩和细节特征。合成结果采取了均值降噪。An algorithm for multi-meanings image synthesis which based on wavelet transform is presented.Firstly,the two target images were decomposed by wavelet transform with the scale which was selected by visual angles.Secondly,the multi-meanings image viewed in two angles was reconstructed by elective fusion algorithms and variable weight.The algorithm based on wavelet transform and two existing algorithms were compared and analyzed.Experimental results show that the images produced by this algorithm had better characteristics of multi-meanings and compelling visible effect,and contained sufficient color and details of original images.Finally,the results were averaged for denoising.

关 键 词：计算机图像处理 小波变换 视觉特性 图像融合 多义图像 目标图像 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]