Multi-spectral and SAR images fusion via Mallat and À trous wavelet transform

The information which is contained in the multi-spectral and SAR images have different characteristic. Multi-spectral images contain a great deal of spectral information, whereas SAR images contain rich texture information, such as buildings and road network. SAR and TM images fusion based on the wavelet transform ensure the fusion image showing more spatial detail, not only conserving the spectral information of the multi-spectral images and reducing the distortions as well, but also highlighting the texture information of the SAR image. In this paper, the wavelet transform-based image fusion methods by using SAR and TM multi-spectral images are implemented by Mallat and à trous algorithms separately. Before the image fusion, both of the SAR and TM images have to be geographic coordinate registered in order to they having the same pixel size. In wavelet decomposition, the decomposition level is determined by statistical of entropy value. According to SAR and TM image fusion based on wavelet transform, it can be seen that the fusion image is greatly improved and both of spectral and textural information are enhanced. The value of entropy, variance, average gradient and correlation coefficients of the fusion image are analyzed for two different algorithms evaluation. By analyzing the results, It can be concluded that the image fusion by à trous wavelet transform has a good effect in experiment.