一种使用曲波变换的多聚焦图像融合方法  

Multi-focus Image Fusion Scheme Based on Curvelet Transform

作　　者：储彬彬[1] 何晓卒[1] 路刚[2] 漆德宁[1] 

机构地区：[1]解放军炮兵学院信息工程教研室,安徽合肥230031 [2]解放军炮兵学院计算机教研室,安徽合肥230031

出　　处：《电子质量》2009年第8期4-6,9,共4页Electronics Quality

摘　　要：提出一种使用曲线波(Curvelet)变换的多聚焦图像融合方法:首先采用非均匀采样的快速傅里叶变换(FDCT_USFFT)实现方法的二代Curvelet变换,对待融合图像进行分解;然后对不同尺度和方向上的曲波系数采用设计的融合策略进行融合;最后通过融合系数重构得到结果。实验结果表明该方法在主观感觉和客观评价指标上较小波分析的融合算法均有一定的提高。A multi-focus image fusion scheme based on Curvelet transform is put forward in this paper. First, the input multi-focus images are decomposed by Curvelet transform, which is realized by Unequally-Spaced Fast Fourier Transforms. Second, the decomposed coefficients in different scales and directions, are fused by corresponding fusion rules. The fusion image is acquired at last by composing the fused coefficients. Experiments show that this scheme have acquired well effects both in the subjective vision aspect and the objective estimation parameters.

关 键 词：图像融合 多尺度几何分析 曲线波变换 小波变换 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]