Color Image Enhancement Based on Multi Scale Retinex and DWT for Low Contrast Noisy Image

Abstract

In the field of image enhancement, camera captured images are low contrast images under the condition of insufficient light, and corrupted by noise during processing, so it is difficult to identify or distinguish the content of the original images. Well known Multi-scale retinex algorithm (MSR) is used to reduce the low contrast effect on images caused by improper lightening condition. But this algorithm causes strong saturation in colors of color images and also could not overcome effect of noise that do not fit for human vision. This paper presents a concept of Discrete Wavelet transform (DWT) based retinex algorithm. In order to reduce strong saturation and effect of noise, DWT-MSR is quite efficient algorithm. Experimental results show that observations are better than conventional MSR algorithm. Algorithm is being tested for very low contrast images and performs better. This paper concludes that this algorithm may be efficiently used in the field of image enhancement.