A new technology of remote sensing image fusion

Abstract

Wavelet packet transform stands out in the field of image fusion for its good frequency characteristics, and pulse coupled neural network (PCNN) has a unique advantage in image processing. To resolve the problem of multi-spectral remote sensing image fusion, in this paper, we put forward an algorithm combined the wavelet packet and PCNN based on HIS transform.The algorithm will be carried out as follows. Firstly, the TM images will be converted into HIS space, and then the luminance component and the high-resolution image will be broken into multi-scale by wavelet packet. Secondly, according to the frequency domain characteristics of the wavelet packet decomposition, we respectively use a method of weighted average in the low-frequency domain and a method of PCNN in the high frequency domain to select reconstruction coefficient.We can get a fused luminance component by taking inverse wavelet packet transform to be reconstructed. Finally, we can obtain the fusion image by taking inverse HIS transform. The experimental results show that the algorithm can be not only to retain the spectral information, but also greatly improve the spatial resolution of multispectral images, has a good fusion effect.