研究不同小波基函数下图像融合效果的差异性,结合实验结果、图像特性以及小波基函数的性能对小波函数进行筛选,最后确定以sym4小波函数作为图像处理的小波变换基函数；以融合图像低频部分的空间分辨率为评价对象、以熵差和互信息量为评价指标,对不同层次的小波分解融合图像的性能做了系统分析,分析表明,当小波分解为4层时,指标均达到最优化；重点研究局部方差准则、局部能量准则两种融合规则,最后提出基于局部相关系数方法的融合规则,以熵差、交互信息量、联合熵、均方根误差、标准差等多个指标作为评价指标,分别对三种规则下图像融合的效果和质量,进行了客观评价,评价结果表明,基于局部相关系数方法较基于局部标准差准则和基于局部能量准则有优势.%In this paper, we study the difference of image fusion effect with various wavelet basis functions. Combining the experiment result, the image characteristics and the performance of wavelet basis function, we screen the wavelet functions and finally select sym4 as the wavelet transform basis function for image processing. The spatial resolution of low-frequency part of the fused image is taken as the evaluation object, the entropy difference and mutual information are taken as the evaluation parameters, we carry out system analysis on the performances of the fused images in different wavelet decomposition level. The analysis results indicate that when the wavelet is decomposed to 4 levels, all the evaluation parameters achieve their best results; two fusion criteria, the local standard deviation criterion and the local energy criterion, are elaborately studied in this paper. At last, we propose a new fusion criterion which is based on local correlation coefficient, that is, multiple parameters such as the entropy difference, mutual information, joint entropy, mean-root-square error and standard deviation, etc. are used as the evaluation parameters. Objective evaluations on effect and quality of image fusion are carried out with three criteria respectively, the evaluation results show that the new criterion is in predominance than the local standard deviation-based criterion and the local energy-based criterion.
展开▼
