A new image fusion technology based on object extraction and NSCT

Abstract

In this effort, we proposed an new image fusion technique, utilizing Renyi entropy’s object extraction and Non-Subsampled Contourlet Transform (NSCT), for improved visible effect of the image. NSCT is a multiscale transform method, it is a shift-invariant, linear phase, ‘‘true” two-dimensional transform that can decomposes an image into any directional sub-images to capture the intrinsic geometrical structure. In this paper we decompose visible image into 21, 22, and 23 directional sub-images at three different level respectively. Image enhancement is performed at the decomposition level and fused. Renyi entropy is a generalized information entropy. Infrared image can be divided into two parts of the object and the background through the maximum value of Renyi entropy. Image fusion is performed after NSCT and Renyi entropy. The fused image has significantly improved brightness and higher contrast than other images. In order to evaluate the proposed method, information entropy (IE), standard deviation (STD), spatial frequency (SF) and mutual information (MI) are adopted to compare with Laplace, wavelet, and NSCT et al. Results are shown that all evaluation value of the proposed method is higher than that of other methods, and it is a better image fusion method.