A Color Image Enhancement Technique Using Multiscale Retinex

Abstract

In this work we develop a new technique to enhance color images using multiscale retinex theory. The proposed technique consists of four major steps. In first step Principal Component Analysis (PCA) is used on the input colored image to obtain the luminance channel. In second step luminance channel is enhanced using multiscale retinex. In third step the ratio of enhanced luminance to the original luminance is calculated. Meanwhile new red, green and blue values of images are computed by multiplying the luminance ratio with initial red, green and blue values of the colored image. In fourth step, contrast stretching is performed on each red, green and blue channel separately. Detailed experiments were performed on 81 colored images having non-uniform luminance. The results of proposed technique show that the brightness and the overall visual appearance of the image are increased as compared to conventional methods such as histogram equalization, histogram specification and gamma correction with less execution time.