Scene simulation under different weathers

Grey level of image is greatly influenced by weather conditions if a TV camera is used out of door. In this paper. relationships between grey level and weather conditions are addressed. First, by using daylight model, daylight distribution of scenes at different time and different weathers can be estimated by using a empiric formula. Subsequently, an image model of daylight is founded after analyzing a great deal of the experimental results. In this model, average grey level of image and its variance in a homogenous region can be predicted if some parameters are determined by the experiments in advance. Finally beginning from the simplification of imaging model, a normalized grey level invariant is derived which can provide a scene simulation approach under different weather conditions if one of the corresponding image of the same scene at one weather is known. The experiment results demonstrate the feasibility of our theoretical analysis and are promising.