In a traffic control system, in order to achieve the data of moving vehicles, hardware is installed to directly measure them in the early days. If we use computer software to calculate the data from traffic video, we can gain more convenience and flexibility. This thesis is focused on the research of the methods of achieving the data of moving vehicles from traffic video by computer.Moving object detection and tracking can be divided into these steps: background extraction, moving blob extraction, moving blob's position achieving and moving object tracking. This thesis shows our experiments and analysis on many algorithms in each of the steps. It also shows the improvement of the algorithms made by us. The experiments and analysis demonstrate that the improved algorithms are better then the original ones.Experiments and analysis also demonstrate that moving object detection and tracking based on color images is better than based on the gray images or edge images.In the end we obtain a whole method of moving object detection and tracking. The robustness and real-time property of the method can reach the acquirement of the real application. |