基于改进l_1范数和稀疏表示的SAR与可见光图像快速融合算法

A fast fusion algorithm of SAR and optical images based on improved l_1 norm and sparse representation

针对多尺度融合算法中合成孔径雷达(SAR)与可见光图像融合结果目标信息易缺失、对比度不高的缺点,提出了一种基于改进l_1范数和稀疏表示的图像融合算法,以有效保留源图像的目标信息。首先对SAR与可见光图像经支持度变换(SVT)分解得到的低频系数进行过完备稀疏融合,采用改进l_1范数的稀疏系数融合规则以有效保留源图像目标信息,并进行零均值化处理提高了算法运行效率,然后利用基于区域能量的高频融合规则,最大化保留边缘纹理等细节信息。实验结果证明了该算法的有效性。

Aiming at the disadvantages of easy-to-loss target information and low contrast images of the fusion images of Synthetic Aperture Radar （SAR） and optical images in multi-scale fusion algorithm, a fast fusion algorithm of SAR and optical images based on improved l1 norm and sparse representation is proposed to retain target information of the source images effectively. Firstly over-complete sparse fusion is done to the low frequency coefficient got in Support Value Transform （SVT） of SAR and optical images sparse coefficient ., and based a fusion rule with improved on l1 norm is used to retain target information of the source images effectively. Then zero mean processing is done to improve the algorithm efficiency. Finally the high frequency fusion rules based on regional energy is used to maximize the details of edge texture. The experiment results demonstrate the effectiveness of the algorithm.