An novel image fusion algorithm of remote polarimetric image based on wavelet transform is proposed. At first, three polarimtric images in each band of two bands are decomposed into images of low frequency band and high frequency band by wavelet transform. The fused low frequency approximate coefficients are obtained with average method, and the fused high frequency detailed coefficients are obtained with different area-based feature selection method by setting each coefficient equaling the corresponding input image wavelet coefficients that has the greatest multi-window local deviation. One image in 555 band and the other in 665 band are obtained. Then, the algorithm decomposes the two images in one band into parts with low frequency band and high frequency band by wavelet transform, it adopt the minimum low frequency image wavelet coefficients as the fused low frequency image wavelet coefficients, and high frequency image fusion rules based on the texture detection algorithm of texture homogeneity measurement are used to adjust high frequency image wavelet coefficients. The low frequency component and orientation detail component of image are fused pertinently, features and details from different images are fused successfully and the quality of these fusion images is compared. The experimental result shows that the fused images present the polarization information, acquire more spectral information and higher contrast, and object detection and recognition can be more easy with fused images.