Research On The Application Of Single Vehicle Image Detection Algorithm Based On Fog And Haze Weather

Traffic monitoring system plays an important role in intelligent transportation system.With the frequent occurrence of haze in China,the quality of vehicle images become much lower.The original intelligent transportation system cannot meet demand of vehicle detection accuracy under this circumstances.In order to solve this problem,this thesis first remove the haze for single image under different kinds of vehicle,then perform vehicle detection.The main works and contributions of this thesis are summarized as follows:1, Firstly,this thesis analyzes the different monitoring levels between the normal weather condition and the haze air.Secondly,we analyzes vehicle images under the general conditions under the haze air.Finally,according to the different vehicle targets,we class the vehicles into different groups,which is the per-process step for haze removal.2,Secondly,this thesis based on the haze air in Zhengzhou,we try to remove the fog in the single image with the combination of both software and hardware.By analyzing state-of-the-arts signal images haze removal algorithms,we found that the commonly used models usually cannot satisfy the demand under all the road situations,especially for single image haze removal problem.In order to solve this problem,we propose a fusion strategy based on fog mask,which can meet the demand for single image haze removal.3,Thirdly,we perform the vehicle detection after the haze removal process.We used the optical flow method to perform vehicle detection.We selected different detection methods based on the type of the vehicle.Finally,experimental results evaluate the superior performance of our method.