Research On Techniques Of Vehicle Moving Target Extraction And Tracking

Vehicles moving target detection and tracking technology is a key and very basic link in intelligent transportation system, which is the necessary premise to realize the vehicle identification, speed detection, traffic flow hierarchies and control, it is a blend of digital image processing, computer vision, pattern recognition, and many other related fields of knowledge.In this paper, the thesis is the study and implementation of the vehicle moving target in the video sequence extraction and tracking technology, vehicle moving target extraction including vehicles moving target detection and target image segmentation,the main content of the paper has the following several aspects:(1) Image preprocessing and background modeling: image preprocessing mainly include filter and morphological algorithms. motion segmentation method, frame difference method and background subtraction method is commonly used in moving target detection, mainly research on the basic principle of the single gaussian background model and gaussian mixture background model based on background subtraction, the detailed description of the two kinds of algorithm, using the algorithm of gaussian mixture background, detect moving targets, determine the target area.(2) The precise segmentation of the vehicle: moving vehicle accurately segmentation is very basic and critical in intelligent transportation system. Firstly introduces the traditional vehicle image segmentation method, then introduces a kind of based on PDE(partial differential equation) image segmentation method, the experimental results show that accuracy of this method is higher compared with the traditional segmentation method of segmentation. Secondly introduced the GAC active contour model image segmentation algorithm based on PDE(partial differential equation), then aiming at the existence problems of the original algorithm trapped in local minimum value, evolution and initialization problems, respectively from the two aspects avoid falling into local minimum to achieve accurate segmentation and rejoin energy constraints, proposed an improved fast segmentation algorithm and from two aspects of algorithm convergence speed and convergence ability compared and analysed improved algorithm with before one. The experimental results show that the improved algorithm compared with the original algorithm, the image segmentation evolution has not only higher precision, speed, but greatly reduces the computational complexity, improves the anti-noise ability, has the very high application value.(3) The vehicle moving target extraction and tracking: The improved GAC model will be combined with a background subtraction algorithm, this algorithm was designed and implemented a precise motion vehicle segmentation method. Application of active contour and level set image segmentation method in the field of moving target detection and tracking has been widely research and application, this paper adopts a kind of target contour tracking algorithm based on the combination of GAC model and kalman filter method, with a high precision tracking vehicles moving target. The algorithm has high feasibility.