Research And Application Of Image Enhancement Algorithms Based On Retinex

Image enhancement technology is a kind of image processing technology which is widely applied in many domains, and has long been a research focus in the image processing. Retinex theory is a kind of image enhancement which is based on scientific experiments and scientific analysis on the basis of Human Visual System. Its essential idea is to estimate the incident light components from the original image, then removing the component of the incident light from the original image, so that we can obtain the reflection components that response the intrinsic attributes of objects in the original image. And the reflection image is the natural shape of the image. Retinex algorithm may be in balance in dynamic range compression, edge enhancement and color constancy, which make it different from the conventional image enhancement algorithms, so Retinex can enhance a variety of images adaptively. Retinex has so many good properties, so Retinex algorithm has been widely used in many domains.According to the difference to estimate the incident image, Retinex algorithms can be divided into two parts:the iterative Retinex and the center/surround Retinex. As the center/surround Retinex is effective and can be simply realized, we mainly embark from the center/surround Retinex in this paper. And then we study several classic image enhancement algorithms, and analyze the advantages and disadvantages of these algorithms. However, due to its own shortcomings of Retinex algorithm, the enhanced image will produce halo and amplified noise. In this paper, we propose a new method of Retinex enhancement to improve those problems. By comparing the experimental results, it proves that the new algorithm can not only improve the quality of the image, but also to remove the halo, and suppress the noise. Subsequently, we apply the Retinex algorithm to low-contrast foggy images in this paper. Experiments show that Retinex algorithm not only can achieve a clear picture of the fog, but also can keep the color of the original image.