Study On The Restoration Of Image And Video Under Fog And Rain Conditions

Bad weather;like fog, rain, snow;makes many troubles to people's daily life and work. So the study on restoration of image and video under fog and rain conditions has great applications in many fields such as: target recognition, remote sensing, traffic navigation and martial defense and so on. And many researchers pay much attention to this technique nowadays.In this paper, we propose several algorithms to restore and generate images and videos in fog and rain, which is based on the physical mechanisms. Generally, our main work is as follows:In the first chapter, we introduced some concepts of digital image processing and several classic algorithms of image restoration. Furthermore, we summarized the researches about the restoration of images and videos in fog and rain, both foreign and domestic.In the second chapter, we propose a simple method to remove weather effects using only a single image, which need few interactions to get much better results. In the algorithm, We fist propose a new primary model for de-weathering with only one parameter which is simple and easy to manipulate. Then we make full use of the information of the image itself to build a new model to estimate the opposite depth of the scenery. Finally, we propose an interactive post modifying algorithm to adjust the local restoration effect finely, which can be use selectively to get much more effect.The common disadvantage of many physical restoration algorithms is that they had to try many parameters to get the better results. In our third chapter, we propose a new efficient optimal restoration algorithm for foggy images, which based on both degrade mechanisms and the minimum distortion rule in information theory. We use the property of projective distortion to acquire the optimal parameters which meets the theory of minimum distortion rule. Then we get the optimal estimate of original signal. Finally we successfully de-fog and recover the degraded images automatically.The generation and restoration technique on videos which captures in rain is a funny problem and will be much more useful. In the fourth chapter, we first proposed an algorithm togenerate video of dynamic rain scenery base on a single image. Then we proposed a simple algorithm to remove rain from videos by analyzing the statistical information of the drops and the pixel intensity, which is suitable for the videos with static background.Finally, we sum up all researches in this paper, and give the direction of following research.