| In order to manager the traffic system, the traffic video monitoring system is used more commonly. As the traffic video processing technology is the basic technology, more and more researchers focus on it. However, the traffic video images are captured by cameras installing on the road, which are often effected by the changeable outdoor weather. Now, most areas in China are facing serious hazy problem. Hazy weather conditions may cause traffic video image degradation, decrease the contrast of the image, and make the images fuzzy. It will lead to the problems such as mistaken result for vehicle identification and tracking, and ultimately affect the effective operation of the whole traffic video processing system. Therefore, the research for traffic video dehazing is the key to traffic video processing in hazy weather.The existing dehazing algorithms have three problems: having poor real-time performance, contrast distortion after restoration, cannot eliminating the influence of dense haze. In this paper, we analyse the characteristics of the traffic video and focus on the study of traffic video image dehazing. The main research contents include:(1) In order to solve the problem that image dehazing algorithms have poor adaptability, which will occur contrast distortion after restoration or cannot eliminating the influence of dense haze, an adaptive contrast-enhanced image dehazing algorithm, which is based on the image-block-contrast-enhanced method, is proposed. This algorithm estimates the initial transmission value through contrast and the amount of haze in the image, which make it strongly adapt to images with different amounts of haze.(2) A traffic video dehazing method based on the spatial and temporal correlation is proposed to improve the processing speed for traffic video dehazing. This method based on the adaptive contrast-enhanced image dehazing algorithm, calculates the parameters periodically by the time continuity, reduces the image area for dehazing by lane detection. Further, it takes advantage of the spatial distribution of cameras and speeds up the processing for video dehazing in multiple cameras.(3) Combining with the characteristics of the traffic video processing applications, study the workflow for traffic video processing in hazy weather, and implement traffic video dehazing system. This system can realize multiple traffic videos dehazing and output dehazing results, having good applicability.This paper studies two aspects of dehazing which are image dehazing and traffic video dehazing. And finally realize a traffic video dehazing system to support the traffic video processing under hazy weather. The proposed image dehazing algorithm focuses on the problem that the existing algorithms have poor adaptability for images with different amounts of haze. While the proposed traffic video dehazing method mainly improves the processing speed and ensures the traffic video real-time processing requirements. Through proposing a traffic video dehazing system, our algorithm can directly be used in applications. |
PDF Full Text Request