Curvelet变换在多聚焦图像融合中的应用  

Application of Curvelet transform on multifocus image fusion

作　　者：安红岩[1] 张正肖[2] 杨武年[1] 

机构地区：[1]成都理工大学地质灾害防治与地质环境保护国家重点实验室遥感与GIS研究所,成都610059 [2]成都理工大学信息管理学院,成都610059

出　　处：《计算机工程与应用》2010年第8期170-173,176,共5页Computer Engineering and Applications

基　　金：成都理工大学青年科学基金(No.2008-332)

摘　　要：综合分析Curvelet变换的特性,提出一种基于Curvelet变换的多传感器图像融合算法。首先将图像进行Curvelet变换,然后基于子块区域方差和区域能量对图像的低频系数和高频系数采用不同的融合规则将Curvelet系数融合,最后进行重构得到融合结果。对该方法得到的融合图像进行了主客观评价和对比。实验结果表明,该方法得到的融合图像在图像边缘等细节上比传统方法和小波算法得到的图像具有更好的视觉效果,在均方根误差、偏差指数、峰值信噪比等客观指标上都优于其他方法。The characteristics of the Curvelet transform are studied,and a new algorithm to fuse multi-sensor images based on the Curvelet transform is proposed.Firstly,the source images are decomposed using Curvelet transform,then the low frequency and high frequency coefficients are fused with different fusion regulars based on Area Standard Deviation and Area Energy.Finally the fused coefficients are reconstructed to obtain fusion resuh.The fusion image obtained is evaluated in subjective and objective factors.Compared with the traditional algorithm and wavelet algorithm,experimental results show that this method yields fusion image with better visual quality and better objective indexes in Root Mean Square Error（RMSE）,Difference Coefficient（DC）,Peak Signal-to-Noise Ratio（PSNR）.

关 键 词：多聚焦图像 CURVELET变换 图像融合 融合规则 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]