Research On Image Defogging Enhancement Algorithm Based On Retinex Theory

With the development of image processing and computer vision technology, more andmore outdoor visual systems are used in our daily life. However, many computer visualsystems are very sensitive to the weather. Because of the atmosphere scattering phenomenon,weather degraded images captured in fog suffer from poor contrast and low visibility. Thissignificantly affects the function of traffic systems and outdoor visual systems. In order toimprove the robustness and reliability of outdoor visual systems, it is meanness to enhancethe definition of the subjects’ details.Images defogging techniques fall into two categories. One is based on atmosphericscattering physical model, the other one is based on image enhancement. Under the basis ofanalyzing the fog imaging mechanism, we research the problem of foggy images’ clearness.The main researches as follows:(1) Images defogging method based on atmospheric scattering physical model.In this paper, images defogging method based on dark channel priority is studied as aneffective method belonging to image defogging method based on atmospheric scatteringphysical model. We introduce tolerance mechanisms to improve the traditional algorithm.The improved algorithm can handle the bright areas effectively, correct the transmission andavoid color distortion in bright areas. Experiment results have shown that this algorithm caneffectively improve the degradation of foggy and hazy image and enhance its definition.(2) Image defogging method based on image enhancement.In this paper, many images defogging techniques are deeply studied, such as histogramequalization, homomorphic filtering, wavelet enhancement and retinex algorithm.Considering that multi-resolution wavelet analysis can both describe the flat areas anddetails in the image, retinex algorithm can improve the image’s brightness and contrast, awavelet-based retinex algorithm for image defogging is proposed, the traditional retinexalgorithm is improved by using bilateral filter. We analyze the experiment results bysubjectively and objectively assessing image quality, Experiment results have shown thatthis algorithm can effectively handle different foggy images, they also validate itseffectiveness and real-time.