The Research Of Traffic Parameters Acquisition Algorithm Based On Video

With the growing traffic pressure, the traditional traffic information parameters acquisition technology cannot meet the necessary of actual demand. The video-based detection technology has become one of the most effective ways of traffic information parameters extraction.In this paper, we take the traffic scene video images taken by fixed cameras as the research object, with the aim of getting the traffic information parameters in the scene, at the base of background extraction、target detection、segmentation、removal of the shadow and target tracking for the traffic scene video images, to accomplish the acquisition of traffic information parameters.The main work of this thesis can be summarized as follows:Firstly, we proposed the diagram of traffic information parameters acquisition algorithm and each part of it was briefly explained as well, then introduced some basic knowledge of digital image preprocessing.In the second place, we has introduced and analyzed the key technology of ex-tracting moving vehicle with emphasis on background extraction and adaptation, and the statistic histogram method therein was improved. On the one hand, the calculated amount of the improved method is greatly reduced to fit the needs of real-time process-ing; on the other hand, this method achieves better effect than other several methods when there are many vehicles or vehicles travel more slowly.Then we discussed some commonly used shadow removal algorithm in de-tail, and corresponding MATLAB programs were written to analyze their effects. The original algorithm of shadow removal using edge information reconstructed the moving object by the way of the level scanning and vertical scanning. The shadow cannot be cleaned because of space position occlusion when there are many vehicles. According to this, we proposed a new algorithm for the part of reconstruction of the moving ob-ject, which used the skipping window to determine the adscription of pixels of differ-ent vehicles. The experimental results verified that this new algorithm achieves better effect than the original algorithm of shadow removal.At last, we introduced the key acquisition algorithm of vehicle flux and speed:Virtual coil method and Vehicle tracking method. After setting the Virtual coil, we calculated the ratio between the number of pixels which belong to the ve-hicle in the coil and the area of the coil, then compared the ratio with predefined threshold, to determine whether a vehicle pass or not. And we avoided cleverly to be repeat count for the vehicle by setting two state parameters. Against vehicle fracture problems, we proposed a algorithm of Vehicle Merging according to its space infor- mation, which merged successfully the parts of the vehicle broke away together to the corresponding same vehicle, thus the error of Vehicle tracking was greatly reduced.