非线性Contourlet变换和区域特征匹配的图像融合  

Image fusion algorithm based on nonlinear Contourlet transform and regional feature matching

作　　者：蒋炯辉[1] 张纯容[2] 黄辉[3] 刘刚[4] 

机构地区：[1]浙江工业大学之江学院,浙江杭州310024 [2]成都航空职业技术学院,四川成都610100 [3]温州大学,浙江温州325035 [4]浙江工业大学,浙江杭州310024

出　　处：《传感器与微系统》2017年第2期53-56,60,共5页Transducer and Microsystem Technologies

基　　金：浙江省自然科学基金资助项目(LY15F020033)

摘　　要：针对图像融合中存在边缘轮廓表示不清晰和计算冗余度过高的问题。提出了将非线性估计Contourlet变换和图像区域特征相结合的融合算法。通过非线性估计保留Contourlet高频稀疏矩阵最重要系数频带,对低频区域、高频区域采用不同的区域特征策略进行图像融合。实验结果表明:该图像融合算法不仅可以提高图像融合的视觉效果和图像质量、而且可以去噪音,增强图像。Aiming at problem of unclear edge contour and high computing redundancy exists in image fusion,put forward image fusion algorithm combined with nonlinear approximation contourlet transform and regional feature. By nonlinear estimation keeps the most significant coefficients frequency band of Contourlet high-frequency sparse matrix. Fuse image by using different regional feature strategies to low-frequency region and high-frequency region. Experimental results show that this image fusion algorithm can not only improve visual effect and image quality, but also can remove the noise and enhance the image.

关 键 词：图像融合 CONTOURLET变换 特征匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]