基于像素分类的自适应SAR图像融合算法*

作者：高佳城，林岳松，陈华杰 日期：2009-04-21/span> 浏览：3879 查看PDF文档

基于像素分类的自适应SAR图像融合算法*

高佳城，林岳松，陈华杰
（杭州电子科技大学 信息与控制研究所，浙江 杭州 310037）

摘要：提高纹理清晰度、保护边缘信息是合成孔径雷达（SAR)图像融合的重要目标。针对该问题，提出了一种基于像素分类的自适应SAR图像融合算法。首先使用canny算子提取图像的边缘并分类，然后利用混合高斯模型和隐马尔可夫树模型对小波系数进行建模；在此基础上使用EM算法求得模型参数，并进一步得到隐状态的概率，也就确定了小波系数的混合高斯分布；接着对两个待融合小波系数不同的类型组合采用不同的融合策略，并以隐状态概率加权；最后通过小波反变换、边缘分类增强获得融合以后的图像。实验结果表明，和传统的融合算法相比，该算法取得了更好的融合效果。
关键词：像素分类；自适应图像融合；混合高斯模型；隐马尔可夫树模型；合成孔径雷达（SAR）
中图分类号：TP751.1文献标识码：A文章编号：1001-4551(2009)03-0016-04

Adaptive SAR image fusion algorithm based on pixel classification
GAO Jiacheng, LIN Yuesong, CHEN Huajie
(Institute of Information & Control, Hangzhou Dianzi University, Hangzhou 310037, China)
Abstract: Improving texture and preserving edge is the important target of synthetic aperture radar(SAR) image fusion. Aiming at the problem, adaptive SAR image fusion algorithm based on pixel classification was proposed. The edges of two source images were firstly distilled using canny operator, followed by edge classification. Then mixture Gaussian model and hidden Markov tree model were used for image modeling of wavelet coefficients. The model parameters were computed using EM algorithm, which was followed by the computation of hidden states probabilities. After having built the mixture Gaussian models of wavelet coefficients, the fusion rules were selected in terms of the distribution of wavelet coefficients, which was followed by linear compages based on probabilities of hidden states. The final fusion image was obtained by wavelet inverse transform and edge enhancing based on the edge classification. Experimental results show that the algorithm provides significant improvement over conventional image fusion methods.
Key words: pixel classification; adaptive image fusion; mixture Gaussian model; hidden Markov tree model; synthetic aperture radar(SAR)
参考文献(References):
［1］SHAHID M, GUPTA S. Image Fusion across Bands［C］// Proceedings of the Eighth International Symposium on Signal Processing and Its Applications. Sydney:［s.n.］,2005:811-814.
［2］CHIBANI Y,HOUACINE A. On the Use of the Redundant Wavelet Transform for Multisensor Image Fusion［C］// The 7th IEEE International Conference on Electronics, Circuits and Systems. Jounith: [s.n.],2000:442-445.
［3］LI H, MUNJANATH B, MITRA S. Multisensor image fusion using the wavelet transform［J］. Graph. Models Image Process,1995,57(3):235-245.
［4］BURT P. The Pyramid as a Structure for Efficient Computation, Multiresolution Image Processing and Analysis［M］. London:SpringerVerlag,1984.
［5］王宏，敬忠良，李建勋. 多分辨率图像融合的研究与发展［J］. 控制理论与应用，2004，21（1）：145-151.
［6］BURT P J, KOLCZYNSKI R J. Enhanced Image Capture through Fusion［C］//Proceedings of Fourth International Conference on Computer Vision. Berlin: [s.n.],1993:173-182.
［7］蒲恬，方庆矗吖.基于对比度的多分辨图像融合［J］.电子学报，2000，28（12）：116-118.
［8］CROUSE M S, NOWAK R D, BARANIUK R G. Waveletbased statistical signal processing using hidden markov models［J］.IEEE Transactions on Signal Processing,1998,46(4):886-902.
［9］ROMBERG J K, CHOI H, BARANIUK R G. Bayesian treestructured image modeling using waveletdomain hidden markov models［J］.IEEE Transactions on Image Processing,2001,10(7):1056-1068.
［10］于秋则. 合成孔径雷达（SAR）图像匹配导航技术研究［D］.武汉：华中科技大学图像识别与人工智能研究所，2004.
［11］覃征,鲍复民,李爱国,等.多传感器图像融合及其应用综述［J］.微电子学与计算机，2004，21（2）：1-5.
［12］顾国松，林岳松，陈华杰.UML和ICONIX在SAR图像融合平台建模中的应用［J］.机电工程,2008,25(2):31-34.