A LUMINANCE BASED COLOR IMAGE ENHANCEMENT METHOD USING MULTISCALE RETINEX

Abstract

Retinex theory is used to enhance images under varying illuminating conditions. It has wide range of applications, especially in medical imaging such as, Computer Tomography (CT), Magnetic Resonance Imaging (MRI), X-Rays, etc. It can also be used for security and safety purposes like landing of planes in the foggy environment. It can be used to improve the efficiency of algorithms in multimedia monitoring / surveillance, vehicle detection and face detection. Details of the image are more clear after the retinex enhancement algorithms. Different implementations of the retinex based algorithms have been published in image processing fields.

In this work I have developed a new algorithm based on luminance using multiscale retinex to enhance color images. The proposed system consists of following major steps. Firstly, Principal Component Analysis (PCA) transform is used to obtain the luminance channel. PCA provides orthogonality between channels, thus allow the colors to remain stable despite the modification of the luminance. Secondly, luminance channel is enhanced using multiscale retinex. Then the ratio of enhanced luminance to the original luminance is calculated. New R, G and B values of image are calculated by multiplying the luminance ratio with initial r, g, b values of the image. At the last step, contrast stretching is performed on each r, g, b channel separately. The results of the proposed method show that the brightness and the overall visual appearance of the image are increased.