A Vehicle Tracing Approach Based on Video and Road Network

Route backtracking has been applied in many fields, especially in tracing criminal cars. However, because of weak light, shading or shelter, plate numbers are unable to be recognized and the vehicle cannot be traced. So, this paper put forwards a route backtracking method based on video data and road network. the method get routes information by means of vehicles features extraction and matching method with time and spatial constraints. Time constraints include time records in each video and time range that car cost on every road, and spatial constraints include connectivity and topology of roads. Meanwhile, there are also vehicle constraints, such as vehicle´s outline, color, type and plate number. and this paper uses Mean Shift algorithm to extract and match the candidate vehicles and target vehicle. Finally, an experiment is put forward to prove the availability and efficiency of the method.