Research And Implementation Of The Key Technologies Of Video Processing In Intelligent Traffic Under Multiple Scenes And Complex Environments

Video processing in intelligent transportation technology is the important technology to deal with the traffic problem today.The technology is implemented by processing video streaming,analysising various video traffic behavior,and then deciding behavior and giving the response.It relates to computer vision,machine learning,pattern recognition,and other areas of the frontier technology.In this paper,we tell the intelligent traffic video processing method in a variety of complex conditions to remove interference,and the decision method for a variety of scenarios of traffic behavior,and the intelligent transportation system framework of video monitoring and adaptive technology.This paper gives method to detect and eliminate the shadow under the light in image.Based on law of RGB nature in background and prospects,giving the method to detect and eliminate the shadow;Based on the chromatic aberration properties in the background image and real-time image,HSI color space model of the shadow detection method is given,to eliminate the shadow effect;Based on the characteristics of the shadow on the geometric properties,a shadow line searching method to detect and eliminate the shadow is given.The stability of image environment is studied,and several methods to remove image noise are given.Based on stable Jittering and stable shaking proportion of the leaves in a fixed position,jitter detection method is given,to eliminate jitter interference.Based on noise points not getting through in image interest area,the interested region invasion method is designed,to eliminate the random noise.To reduce the impact of cloud changes leading to strong changes in image brightness,the method of adaptive digital aperture to adapt to changes in illumination is used.Detection and decision method is given for different scenarios of traffic behavior.Based on forbid area model,illegally forbid area determination methods is given,and get the vehicle information of Illegally parking and capture it.According to the nature characteristics of road litter size,litter judgment method is given,and to detect litter and give alarm prompt.Based on the brightness and time characteristics of electrical sparkle,electrical sparkle detection algorithm is given,and then detects events of electrical sparkle in conductor with rail train.In this paper,the basic framework and adaptive methods of the intelligent traffic video capture system are given.It realizes the function of video stream acquisition,image processing,the judgment of illegal behavior,the capture of illegal evidence and the network transmission.Based on fractional and polynomial prediction model,the adaptive prediction method was designed;the vehicle trajectory prediction technology is improved.Based on vehicle trajectory statistical data,prediction and triggering coordinates of the vehicle are limited,thus reducing the error rate.Against the wrong trigger problem because of larger vehicle detection box,box narrowed methods by searching vehicle bottom is designed,to reduce misjudgment rate.