基于非下采样Contourlet的熵测度图像融合算法  

An Image Fusion Algorithm with Entropy Measure Based on Non-subsampled Contourlet

作　　者：李骜 李一兵[1] 刘丹丹[2] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001 [2]黑龙江科技学院现代制造加工中心,黑龙江哈尔滨150001

出　　处：《兵工学报》2011年第11期1348-1352,共5页Acta Armamentarii

基　　金：国家自然科学基金资助项目(50904025);船舶工业国防科技预研项目(10J3.16)

摘　　要：活性测度是图像融合过程中的重要衡量工具。通过其提取的某种特征信息,来决定哪幅输入图像的特征更为明显。在基于多分辨率分解的融合算法中,常用的活性测度只考虑各层高频子带系数本身,忽略了低频系数所提供的信息。考虑到综合高、低频系数对于活性测度的影响,基于非下采样Contourlet变换良好的平移不变性、多方向等特性,提出了一种在一般方法的基础上,添加由低频系数所得的掩模,再作为活性测度的融合算法。给出了不同活性测度的融合图像以及客观性能评价指标。结果表明,熵掩模测度的融合结果优于其他几种传统测度。The feature information extracted by using activity measure,a very important measure in the process of image fusion,determine a certain input image with more obvious characteristics.In wavelet-based fusion algorithm,the common activity measures consider the coefficients of all high-frequency sub-bands themselves only,and ignore the information provided by low-frequency coefficients.An algorithm based on the general method and added the entropy-masking obtained from the low-frequency coefficients as an active measure is proposed in this paper to comprehensively consider the impact of the high and low frequency coefficients on the activity measures.And,the fused image with different activity measures and objective performance evaluation index are provided also.The results show that the entropy-masking measure is better than several other traditional measures for image fusion.

关 键 词：信息处理技术 熵 活性测度 图像融合 非下采样Contourlet 多分辨率分析 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]