| Fog images and videos are poor contrast and low visibility because the particles in fog atmosphere absorb and scatter light, which affects the traffic systems and outdoor visual systems. Hence, it is significant to enhance the low-contrast images and videos which were degraded by fog. At present, researches about the visibility enhancement for fog-degraded images, especially videos, are still under discussion. Due to the shortages of current visibility methods, such as complexity, low operation speed, gray images only etc., many methods are restricted in application of real-time image operation.For the purpose of real-time visibility enhancement without complex sensors, this paper studies visibility enhancement methods from two aspects: contrast enhancement and restoration methods based on fog degraded physical model.On the aspect of contrast enhancement, this paper studies the methods of histogram equalization with global, local and interpolating adaptation; homomorphic filtering on frequency domain and Retinex algorithm on human vision model. Improved methods were proposed to meet the acquirement of real-time application based on the researches of SSR and MSR.On the aspect of restoration methods based on fog degraded physical model, this paper analyses the atmospheric scattering physical model, studies the previous research productions, and proposed a visibility enhancement methods which doesn't require complex sensors and user interference but solely relies on single image information. The proposed method can be used in real-time operation and has quick operation speed.In this paper, real-time visibility enhancement methods for fog degraded images and videos are proposed based on the study of contrast enhancement algorithm and model restoration algorithm, which can be a basic work for further research in this area. Moreover, the methods in this paper are aimed at color images and videos which will make the methods for wider application. |
PDF Full Text Request