COLOR CORRECTION BASED COLOR IMAGE ENHANCEMENT

Abstract

Color correction is an important step in image enhancement. It improves the appearance of the images affected by color cast. There are many color correction algorithms available in literature. Gray world assumption takes into account human visual system and form the basis for many color correction algorithms. A new image enhancement approach for color correction which modifies gray world correction algorithm is proposed using fuzzy logic technique. Fuzzy logic deals with the uncertainty of information. Here, fuzzy logic is employed to deal with the uncertainty of color cast in the image. RGB channels are fuzzified individually using Gaussian membership function between two sets the LOW and the HIGH. Six fuzzy rules have been defined to find the cast in the color image. Then according to the cast, correction factor is found using the mean of RGB channels as done in gray world based correction. This correction factor is non-linearized by raising it to power of gamma. Out of many performance metric for color correction algorithms, CIE L*a*b metric performance measure called image distance is used in the proposed approach. Image distance has been used as objective function for the optimisation. Bacterial Foraging optimisation is applied for finding the optimal value of gamma which improves the objective function. Results of the proposed scheme are better as compared to Gray World. Quantitative analysis of the results has been done using CIE L*a*b metric called image distance whose value should be lesser for better result.