Improved fusion method for infrared and visible remote sensing imagery using NSCT

This paper introduces the Nonsubsampled Contourlet Transform (NSCT) into infrared and visible remote sensing image fusion and proposes a novel multi-scale analysis method based on region energy. With better direction selectivity and energy convergence compared to that of wavelet, NSCT is suitable for representing detailed information of remote sensing images. In addition, region energy method is better than coefficient mean value in computing low-frequency coefficients. Firstly, the images are decomposed by NSCT. Then, the low-frequency coefficient is calculated by infrared and visible images region energy and the max absolute value is selected as the high-frequency coefficient. The experimental results of Airborne Visual and Infrared Imaging Spectrometer (AVIRIS) and ASTER images show that the proposed fusion method can get more detail information and achieves better visual quality than discrete wavelet based transform and normal NSCT which uses coefficient mean value in low-frequency.