The Image Defogging Algorithm Based On Physics And Nonphysics Model

With the rapid development of information technology, image processing technology has been rapid development and has attracted many attentions of researchers. In recent years, most cities of our country suffered varying degrees of haze weather, the haze weather not only harm to human’s health, but also brings the new challenges of image acquisition for outdoor vision system, the image through computer vision to obtain will have a certain degree of degradation, affect the extraction of right image information.At first, we introduces the research background and significance of fog image processing, fog image enhancement research status is described, then mainly introduced physical model based on the atmospheric scattering and image enhancement algorithm based on the theory of the Retinex. Through the analysis of the formation of fog, light reflection effect, Attenuation model and Air light model, the fog image degradation model is proposed; then the thesis introduces the Retinex Theory, and makes a detailed analysis of several methods: SSR, MSR, MSRCR and SDMSR, last compares the advantages and disadvantages of these methods.Because the image contrast and brightness is not good after the enhancement of MSR algorithm, the defogging algorithm based on the prior dark-channel and theory of retinex is proposed. The proposed algorithm firstly uses the theory of dark channel prior to the defogging image. Then the thesis uses the MSR algorithm for the image enhancement. In the thesis the author uses the mean value, standard deviation,entropy, MSE, PSNR as evaluation standard. The simulation results show that the contrast of hazing image is increased, the average brightness of image is moderate, and the overall effect is good. Furthermore, the clarity of hazy image can be improved in some extent.The traditional Retinex algorithm only considers enhancing the shadows of the image, but without to enhance the brightness of image, and generate the halo phenomenon in the bright place. So the multi-scale retinex algorithm for the foggy image enhancement based on sub-band decomposition is proposed. First, hybrid intensity transfer function was used to take the place of the intensity transfer function of the retinex. This algorithm prevents the saturation phenomenon of the image from the image enhancement. Then, the outputs of the retinex algorithm were decomposed into non-overlapping spectral sub-bands. The simulation results show that the effectiveness and accuracy of the proposed algorithm was verified, and the visual of the image is better, and the contrast of image is also improved. Image distortion degree is small, and the improved algorithm of mist image processing effect is better than that of heavy fog image processing effect.