基于局部特征差异的异源图像融合算法

A multi-sensor image fusion algorithm based on local feature difference

针对现有异源图像融合多以光学图像为主、合成孔径雷达(SAR)图为辅和光学图像极易受传播媒介干扰且不能同时保留纹理细节与颜色信息等问题,提出一种新的基于局部特征差异的异源图像融合算法。首先通过自适应分割将SAR图像划分为规则特征区和不规则特征区两个区域;然后进行平移不变离散小波变换(SIDWT),再根据局部特征差异性设计融合规则,将SAR图像与全色遥感(PAN)图像的小波系数进行融合,以期保留图像的特征信息与色彩信息;最后通过信息量、清晰度等客观评价指标对融合结果进行评价与分析。仿真实验证明了算法的有效性。

The multi-sensor image fusion technology can acquire a more comprehensive,more accurate and more reliable image,in order to understand the scene or recognize the target more easily.However,most existing algorithms are mainly based on optical remote sensing images,supplemented by synthetic aperture radar（SAR）images,but optical remote sensing images are highly susceptible by environmental condition and media interference.The image fusion between SAR images and panchromatic（PAN）images also cannot save the textural feature of irregular area and the color information of regular area at the same time.In view of these problems,a multi-sensor image fusion algorithm based on local feature difference is proposed in this paper.Firstly,the SAR image is divided into regular area and irregular area by a new adaptive segmentation method.Secondly,the feature coefficients of the two areas are extracted using shift invariance discrete wavelet transform（SIDWT）.And according to local feature,a new fusion rule is designed to fuse the corresponding wavelet coefficients of SAR image and PAN image.Finally,the fused image is got after inverse wavelet transform with new wavelet coefficients.Experimental results demonstrate the effectiveness of the proposed algorithm on both the subjective and objective evaluations,such as clarity and information content.