Transport System-based Video Motion Detection

As computing and communications technology is developing rapidly, video surveillance technology has unprecedented breakthroughs, and become more intelligent, especially in the transport system plays an important application, which is used to ensure traffic safety, flow and other aspects. The moving target detection and recognition technology as part of monitoring system technology is critical to make the traditional sense of the artificial control intelligent, and bring post-campaign tracking and analysis the goal of a viable basis. It has been widely distributed for medical image processing, video surveillance, remote sensing data analysis and other fields.This paper is based on TI's TMS320DM6467 production hardware platform to achieve the surveillance video image moving target detection and recognition, and discuss in tail how to build the platform for embedded development and detect moving target, finally identify the vehicle.First describes internal structure and performance of the required various functional modules. then build software development platform, and introduces the upper module and the underlying application modules based on the platform. then focus on the monitoring system's core modules named algorithm modules, mainly for moving target detection and identification. In the moving target detection, The algorithm for moving target detection combine with two methods to improve the timeliness and effectiveness of the algorithm and improved algorithm update method to overcome the potential long-term update distortion phenomenon. After the motion detection processing, use mathematical morphology filtering to optimize the test results for filling the place empty of movement area and isolateing noise points. In the final stage of moving target identification the paper use real-time higher mark line method to detect connected domain and then calculate the connected domain area, set an area threshold to identify the movement of vehicles, and ultimately draw the connectivity of the region's bounding rectangle for meeting the required threshold to indicate the movement of vehicles.The performance of the monitoring system is stable, in the case of clear vehicle space and not overlap of traffic environment, able to adapt to the change of the weather conditions, the size of the deformation and moving objects background color with better real-time detection.