基于视频的车辆检测与跟踪分析-精密仪器及机械专业论文.docxVIP

II II Abstract Intelligent transportation systems (ITS) have attracted considerable research attention in areas such as vehicle detection, recognition, and counting, and traffic parameter estimation. In light of the anticipated availability of low-cost hardware, as well as continuing progress in algorithmic research, computer vision has become a promising base technology for traffic sensing systems. Since vision sensors provide more information than the conventional sensors widely used in ITS, attention is now being focused on vision-based traffic surveillance systems. The main research topic is based on practical application of adaptive of vehicle detection technology, including background modeling, an occlusion vehicle segmentation and vehicle recognition method, a kind of vehicle tracking method. The main content summed up as follows: Firstly, some basic concepts of digital image and processing mode are briefly introduced, then compare and analyze several background modeling method. According to the “ghost” because of the vehicles’ sudden start or stop, presents a homomorphism filtering based background modeling approach。This method is able to not only deal with the illumination change but also successful tested in handle the “ghost” ,and then also can adaptive updating background. Secondly, compared with several models for vehicle recognition, put forward the real-time and complexity, presents a vehicle detection method based on visual length and visual width. This method is mainly to the image after coarse segmentation. In different environment it can be good to detect the vehicles which in monitoring area. Lastly, analyzes some common vehicle tracking method. Because of the computational complexity, those algorithms are difficult to implement, present a vehicle tracking method based on visual length. This algorithm is simple, easy to realize, and can save a lot of time. Key Words：background modeling; vehicle detection; ve