Contrast enhancement for image with non-linear gray transform and wavelet neural network

Abstract

A new contrast enhancement algorithm for image is proposed with non-linear gray transform and wavelet neural network (WNN). In-complete Beta transform (IBT) is used to obtain non-linear gray transform curve. Transform parameters are determined by simulated annealing algorithm (SA) to obtain optimal s space, a new criterion is proposed. Contrast type for original image is determined employing the new criterion. Parameters space is given respectively according to different contrast types, which shrinks parameters space greatly. Thus searching direction and selegray transform parameters. In order to avoid the expensive time for traditional contrast enhancement algorithms, which search optimal gray transform parameters in the whole parameterction of initial values of SA is guided by the new parameter space. In order to calculate IBT in the whole image, a kind of WNN is proposed to approximate the IBT. Experimental results show that the new algorithm is able to adaptively enhance the contrast for image well. the algorithm was large. Existing many enhancement algorithms' intelligence and adaptability are worse and much artificial interference is required. To solve above problems, a new algorithm employing IBT, SA and WNN is proposed. To improve optimization speed and intelligence of algorithm, a new criterion is proposed based on gray level histogram. Contrast type for original image is determined employing the new criterion. Contrast for original image is classified into seven types: particular dark (PD), medium dark (MD), medium dark slightly (MDS), medium bright slightly (MBS), medium bright (MB), particular bright (PB) and good gray level distribution (GGLD). IBT operator transforms original image to a new space. A certain objective function is used to optimize non-linear transform parameters. SA, which was given by William, is used to determine the optimal non-linear transform parameters. In order to reduce the computation burden for calculating IBT, a new kind of WNN is proposed to approximate the IBT in the whole image.