Research Video Detection Technique And Method On Traffic Flow Parameters

At "present, the main characteristic of Chinese city traffic is that one kind hybrid traffic state: Influencing each other among motor vehicle, bicycle, pedestrian, the traffic is confused, traffic block, traffic accident are seen everywhere. Solve the city traffic problem, must grasp various kinds of traffic information accurately. For this reason, we should measure various traffic parameters effectively first. At present, the domestic and international existing vehicle detections, can only measure traffic flow and occupation rate of the motor vehicle, but unable to measure the bicycle and pedestrian effectively. So, in accordance with the main characteristic of Chinese city traffic, and we research the video detection system of the mixing traffic flow that is suitable for national conditions of our country.This thesis relied on province science and technology hall International Cooperation Department in Jilin of《Research image recognition Detection Technique and Method on Traffic Flow Parameters》item and national research development of the point foundation plan(973 plan) of《Transportation in big city City hug foundation science problem research of block up the bottleneck》item in of《City transportation system of organization excellent turn with control》topic. The paper around the technical application about vision that collect of traffic information, and combine the specific research direction of the project, have launched the deep discussion to the expansion of the computer vision in the actual project. The paper begins from modeling of system, and researchs Corner detection, calibration of camera, picture pretreatment, characteristic are drawn systematically on collect of traffic information. We put the groundwork on conner detection techniques, calibration of camera, taxonomic algorithm of goal. Through the study on conner detection techniques, this paper combines operators and advantage of Harris and Forstner , have put forward a kind of new conner detection techniques; and at the calibration of camera, the actual conditions of combining the traffic scene on the basis of analysing that summarizes the standard existing algorithm of the camera have put forward the valid black standard method of case of a kind of convenience; In addition, the paper has also studied the categorised algorithm on the basis of movement goal of supporting the vector quantity machine.On one hand the research of this thesis offers theory support for mixing the traffic flow video , offer basis on which the real-time traffic is controlled with traffic administration, have important theory meanings; Copies of method that literary grace spend on the other hand apply traffic flow parameter measure, offer the effective means for the traffic administration and control, in order to alleviate the traffic jam, improve the utilization ratio of the road, reduce the respects , such as frequency happenning in accident ,etc. , all have important economy and social effect.