基于边缘信息的偏振图像融合算法及评价

偏振遥感图像通常都采用强度、偏振度、偏振角来表征目标偏振特性.本文提出的基于边缘信息的偏振图像融合算法是将三幅偏振图像利用离散小波变换把图像分解成不同尺度的低频和高频部分,采用小波区域窗口和子区域窗口统计把小波系数分类成边缘和非边缘系数,通过这些方法进行有效的边缘细节信息提取.在融合处理中,低频图像的小波系数平均值作为融合后的低频系数,高频细节系数根据不同区域特征选择方法以及对应输入图像小波系数的窗口区域方差来确定融合后高频小波系数.仿真实验结果表明,这样使得融合后的图像细节更真实更丰富,图像的偏振特性体现更为充分,同时减少对源图像的预处理要求,使图像在整体上有较好的视觉效果.从而证明这种方法能够在保留图像微小细节方面获得满意的结果,且算法有效性优于其他的图像融合方法.

Algorithm and evaluation for polarization image fusion based on edge information

An algorithm for polarization image fusion based on edge information is presented, which decomposes three polarization images into low frequency part and high frequency part in different level by discrete wavelet transform. Local window and local sub-window statistic in the wavelet domain is used to further classify the wavelet coefficients into the edge and non-edge coefficients for extracting edge information. In the fusion process, the average of low frequency coefficients is regarded as low frequency coefficient of fused image, the fused detailed coefficients are obtained by several different area-based feature selection methods, and each coefficient equals to the corresponding input image wavelet coefficient that has the greatest multi-window local deviation. The simulation experiment shows that the algorithm can cause the detail of image becoming true and rich, and make polarization characteristic becoming sufficiency after fusion. At the same time, it can reduce pretreatments of original image, and take a great visual effect on the whole. Experimental results show that the proposed algorithm can achieve an excellent effect on retaining subtle image detail. The new image fusion algorithm provides a performance better than other methods.