A Fixed-step Variation Retinex Image Enhancement Based On Wavelet Frame

About seventy percent information that is acquired and transmated by human come from visual information, but there are many uncertain factors exist in the daily life, for example rain and snow, dense fog, dust tornado. The image will be blurred by these factors. And now we should improve the quality of the original image with the help of the image enhancement technique.Nowadays, the new way of image enhancement is Retinex theory and its related algorithm. Color Constancy is the foundation of Retinex therory, the theory is simulate with the humen visual system, restore the lost part of the information because of the external factors, to get the original image restoration.The core of this paper is the research of Retinex algorithm. First of all, make a brief introduction of the traditional Retinex algorithm. Because of wavelet transform has the feature of multiresolution, so we can introduce the wavelet transform into the Retinex algorithm,propose a new kind of Retinex algorithm. The algorithm structure of this paper have these steps. First, convert the original image to HSV space, then apply the logarithmic operations into the V component, do the wavelet transform, use the feature of the wavelet transform, apply the Retinex in each layer to estimate the illumination image, exponent operation, Gamma transform, then change into the RGB space, we can get the enhancement result. The main content of this paper includes the following points:1、 This paper introduces the traditional image enhancement method, in order to get a better ehancement result we usually need to set the certain parameter for the certain image.2、 From the random path retinex algorithm, introduced the several representative Retinex theory developed in recent years. As McCann’s Retinex, center around Retinex, analyzed their advantages and disadvantages.3、 Aiming at the disadvantages of the Retinex algorithm, based on the wavelet multiresolution characteristics, this paper introduced the wavelet transform into the Retinex algorithm. Becaused of calculate the step length is very complicated, reduce the complexity of the algorithm, in this paper we use the fixed step length. The experiment results show that the proposed Retinex algorithm achieves better results than the traditional methods.