Research Article
Bhardwaj J and Nayak A
Abstract
A scheme for multi scale based feature level image fusion for Multimodal medical images has been proposed here. The scheme is based on joint cascaded framework of Lifting wavelet transform (LWT) with Non subsampled contourlet transform (NSCT). These multiscale decomposition (MSD) methods are used back to back to decompose the images of different medical modalities. The low and high frequency coefficients obtained from two stage decompositions are fused according to different fusion rules. At first stage high frequency components (detail information) are fused by Karhunen –Loeve (KL) transform while low frequency by weighted superposition. Later on at second stage the max rule fusion method has been employed for both. Image reconstruction at the intermediary level is obtained by applying Inverse LWT (ILWT) and at final stage Inverse NSCT has been employed. Haar wavelet function is chosen here for its less computational cost property. The effectiveness of the proposed method (LWT-NSCT) is observed by enhanced values of assessment parameters of evaluation indices metric as compared to contemporary and popular transform based fusion methods. Now the attractive features of both transforms i.e. lifting and NSCT like sparse data representation, integer to integer mapping, high vanishing moments and saving of auxiliary memory are attractable features of this method.