Research On Image Enhancement Methods Based On Retinex Theory In A Variational Framework

With the rapid development of computer science and image processing technique, working systems which based on image/video, such as target recognition, automatic navigation, intelligent monitoring, topographic reconnaissance, automatic drive, civilian shooting, etc, have permeate into many research fields and play an important role in national economy and social development. While in the process of image acquisition, the quality of images may degrade by many factors that can affect the extraction of useful information. The degraded quality can be improved by using enhancing technology with some post-processing on the obtained image, which can be helpful to acquire useful information. So there is a significant meaning on both theory and practical application by intensive studying image enhancement.The Retinex theory, which proposed by American physicist E. Land and improved by other researchers, is an image enhancement system. In this paper, the research is about image enhancement method with Retinex theory based on a variational framework. The mainly research achievement and contents are as following:1. In this paper, a novel objective function is built in variational framework based on the prior knowledge of images. Meanwhile an alternating optimization strategy is adopted to solve the proposed model. Different with traditional Retinex algorithms, the proposed method can obtain both reflectance image and illumination image simultaneously. Comparing with other methods by experimental results, the proposed algorithm has a better performance on recovery of original information (such as the edge information or the color information).2. In this paper, a new illumination adjustment algorithm based on image fusion is proposed. The adjusted illumination multiplies with the computed reflectance to obtain the enhanced image. The proposed algorithm not only enhances detail information but also maintains the naturalness and contrast, which improves the subjective visual effect.3. The proposed image enhancement algorithm can deal with many kinds of degraded image, such as low light image enhancement, high dynamic range image compression, color constancy, non-uniform illumination image processing and underwater image enhancement, etc. Experimental results demonstrate the universality of the proposed method.