Road Traffic Status Monitoring

With the rapid development of social development and urbanization,mass social activities gradually increased,such as sports,concerts,religious rallies and so on.Due to different crowed flow with different speed,direction and local crowd density,as well as the objective constraint that the carrying capacity of the road is limited and other factors in similar scenes.Making it is possible,which the crowd prone to congestion and even stampede accident,resulting in loss of life and property in crowd social activities.In order to avoid accidents effectively in group social activities,intelligent monitoring technology has been used in stations,airports,subways and large shopping centers and other fields.While it is the key step to estimate the density of crowd in the intelligent monitoring technology,it is also a research focus on computer vision and image processing.We proposed a method of road traffic status monitoring,which includes motion information detection,motion prospect extraction,motion flow segmentation and flow status monitoring of crowd,for carrying out effective monitoring of population flow status,improving the utilization rate of roads and avoiding unnecessary loss of life and property of citizens.The specific work carried out is as follows:(1)The global optical flow method was applied to obtain the movement information of the scene,including the direction of movement and speed;(2)Extracted the prospect targets depending on the method of particle dynamics;(3)The optical flow angle of the corresponding frame was extracted with the motion foreground as mask,and the dynamic C means clustering algorithm was employed to achieve the prospect motion flow segmentation.(4)Calculated the characteristics of foreground optical flow area and the texture characteristics including energy,entropy,homogeneity and contrast in the optical flow field.Then treated them as the feature vector to establish multiple linear regression model that can estimate the crowd density,and thus achieved the purpose of monitoring the crowd flow.(5)We made experiments by the proposed algorithm and self-organizing neural network to compare the superiority of our algorithm.The experimental results show that our algorithm is feasible and provides some reference value for the follow-up behavior analysis and understanding.