一种基于提升小波变换的快速图像融合方法  

Fast Image Fusion Method Based on Lifting Wavelet Transform

作　　者：李玲玲[1] 丁明跃[1] 周成平[1] 彭晓明[1] 张天序[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所图像信息处理与智能控制教育部重点实验室,湖北武汉430074

出　　处：《小型微型计算机系统》2005年第4期667-670,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金重点项目(60135020FF030405)资助.

摘　　要：目前,多尺度分解的方法已开始应用于图像融合.针对基于传统的多尺度分解的融合方法运算速度慢、对内存的需求量大,不适于实时应用的局限性,提出了一种新的基于提升小波变换的快速图像融合算法.多个源图像分别进行提升小波分解,使用恰当的融合规则合并各尺度对应的分解系数,通过提升小波逆变换得到复合图像.实验结果表明,提出的算法无论在执行时间还是融合后的图像质量上都优于传统的方法,有广泛的应用前景,特别适用于实时系统.Recently, multiresolution techniques especially 2d wavelet transform have been employed for image fusion. Since the traditional wavelet transform is time and memory consuming, which impedes its real-time application, a novel quick image fusion algorithm using lifting wavelet transform is proposed in this paper. First, the source images are decomposed using lifting wavelet transform respectively. Then, the decomposition coefficients are combined by appropriate fusion rules. Finally, the composite image is obtained by using inverse lifting wavelet transform. Experimental results demonstrate that our algorithm is more effective in both computational speed and fused image quality than traditional image fusion methods. It has great significance in practical applications , particularly in real-time system.

关 键 词：图像融合 提升小波变换 多尺度分解 比率低通金字塔 小波变换 

分 类 号：TN911.73[电子电信—通信与信息系统]