Research On Single Image Clearness Technology In Haze Weather

In autum and winter,haze is a common weather phenomena. Influenced by haze,the obtained image is low contrast,not clear and can not reflect the true scence information. Many ofthe features of an object in image is covered by haze and become fuzzy. It has a direct impact on the follow-up work which is based on outdoor vision system.The dark area selection has effects on the accuracy of medium transmission rate estimation. This paper puts forward segmentation based on graph algorithm to confirm the dark areasize. This algorithm first automatically confirm the dark area size by segmentation according to the image scene contents and the change of scene depth,then gain the dark channel in thesegmented dark area size.Avoiding the artificial fixed small aera selection, this algorithm getmore correct medium transmission rate estimation, so it can improve the effect of the final image dehazing. In order to reduce the time-consuming, this algorithm uses guided-filter instead of soft-matting, and adds white-balance step to solve the color mismatched problem.For the further reduction of the computional complexity, this paper studies the real-timedehazing algorithm. This algorithm first calculates the minimum of RGB three channels and then uses edge-preserving filter for filtering. Through this way it could more quickly and accurately estimates the medium transmission rate.The focus of this algorithm is whether filter process can preserve the edge caused by depth changed field. To gain better edge-preserving effect this paper proposes domain transformation quick edge-preserving filter based on the bilateral filter. This proposed real-time dehazing algorithm achieves the good dehazing effect in shorttime.The proposed algorithm is implemented on the personal computer and verifies its properties ofeffectiveness and real-time.