Romote sensing image fusion based on wavelet packet analysis

Wavelet packet algorithm is a method which can provide more precise results for signal frequency band and improve the frequency resolution effectively. Using this algorithm the signal band is divided into multi-level and the high frequency part which is not decomposed by the Wavelet analysis is decomposed further. By analysing both advantages and disadvantages of several traditional remote sensing image fusion methods, in this paper we present a new romote sensing image fusion method based on wavelet packet analysis. For this method, the low frequency coefficients of wavelet packet reconstruction is obtained by using weighted average method on low frequency of the original image's Wavelet packet transform, and the high-frequency coefficient is obtained by using LBP operator on high frequency. Experimental results show that this method improved the time-frequency resolution of the RS image and retained the details of each image's characteristic as much as possible, and this method is better than method which is based on PCA and wavelet transform.