采用WA-WBA与改进INSCT的图像融合算法  

An Image Fusion Algorithm Based on WA-WBA and Improved Non-subsampled Contourlet Transform

作　　者：赵春晖[1] 马丽娟 邵国锋 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,哈尔滨150001

出　　处：《电子与信息学报》2014年第2期304-311,共8页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(61077079);教育部博士点基金(20102304 110013);黑龙江省自然科学基金重点项目(ZD201216);哈尔滨市优秀学科带头人基金(RC2013XK009003)资助课题

摘　　要：在冗余提升变换与多方向分析相结合的图像融合算法改进的非下采样轮廓波变换(INSCT)中,冗余提升部分采用的Neville算子在多尺度分解的过程中丢失了很多方向信息,不利于后续的多方向分析。针对这一问题,该文提出了一组新算子,可以在频带分解部分很好地保持图像的方向信息,并在此基础上提出一种图像融合算法。该算法先用直方图均衡化的方法提高红外图像中目标区域的灰度值;然后用新算子代替Neville算子对可见光图像与处理后的红外图像进行多尺度分解;最后采用基于权重窗口的活动级与加权求和相结合(WA-WBA)的方法替代简单的加权求和方法对低频子带进行融合,采用一致性测度,实现了各子带系数的自适应融合,有效地弥补了基于像素的图像融合方法的不足。实验结果表明,该算法使融合后的图像更好地保持了可见光的细节信息,同时获得了较清晰的红外图像中的目标信息。The Improved Non-Subsampled Contourlet Transform （INSCT） algorithm combines redundancy ascending transform and multi direction analysis. Neville-operator is used in the part of redundancy ascending and direction information is lost in the process of multi-scale decomposition, which is bad for follow-up analysis. To solve this problem, a new set of operators which effectively preserves direction information in frequency band decomposition part is designed in this paper, and a new image fusion algorithm using the operators is proposed. First, the histogram equalization method is used to enhance grey values of the target zone in infrared image. Second, the multi-scale decomposition is performed on visible image and the enhanced infrared image using the new set of operator rather than the Neville-operator. Finally, low-frequency sub-band is fused using a new method incorporating the activity level based on Weighted Average-Window Based Activity measurement （WA-WBA） rather than the simple weighted sum method. The use of neighborhood homogeneous measurement realizes the adaptive fusion of each sub-band coefficient, and the proposed method effectively makes up the disadvantages of pixel-based image fusion method. The experimental results show that the proposed method preserves details of the visible image and obtains clear target information in the infrared image.

关 键 词：图像融合 冗余提升 一致性测度 权重窗口的活动级与加权求和相结合(WA-WBA) 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]