【作者】 谢凌华；

【导师】 沙莎；

【作者基本信息】 中南大学 ， 计算机应用技术， 2008， 硕士

【摘要】 视频监控技术由于检测区域大、系统设置灵活等优点,已成为智能交通系统领域的一个研究热点。针对智能交通系统中的关键技术,研究了基于固定焦距的运动车辆提取算法。针对运动车辆提取的三种常用方法:光流法、背景法和帧间差法,分析了它们的优缺点。在简单情况下,利用序列帧间差法对运动车辆进行了提取,确定了帧间差法的L值,减小了序列帧间差法的空间复杂度。阐述了灰度归类的背景构造方法。在分析背景中静态物体和活动物体的灰度分布后,提出一种使用一个或多个灰度空间表示背景,并利用运动车辆灰度出现的概率小于背景灰度出现的概率的特点对运动车辆进行提取的算法。针对复杂的交通路口场景不能使用单一的算法对运动车辆进行提取的难点,本文使用Robert算子对每帧图像求取边缘信息,利用序列帧差法除去静止物体的边缘从而得到运动物体的边缘信息,并与序列帧差法得到的结果图作差,使得运动车辆灰度块与背景灰度块相分离。通过统计灰度块的边缘信息是否为背景边缘信息对运动车辆区域进行提取。最后,在对已做的工作进行总结的基础上,提出了后续研究工作的思路,对进一步的研究具有一定的指导意义。更多还原

【Abstract】 Due to the wide detecting areas and free system setting of the video surveillance technology, it has become one of the hotspots in the studying field of Intelligent Transportation System. Aiming at the key technology in the Intelligent Transportation System, the essay has made a discussion of the algorithm of extracting the moving vehicles on the basis of the fixed focus.Aiming at three common approaches in extracting the moving vehicles that is optical flow, the background subtraction and coterminous frames differencing, the essay has made an analysis of their advantages and disadvantages. Under some simple situation, by using the sequence coterminous frames differencing, the essay has discussed how to extract the moving the moving vehicles and the result of "L" has been worked out. Therefore, the minimum complexity of sequence coterminous frames differencing space will come out under such simple situation.The essay has expounded the background construction of intensity classification. After analyzing the intensity distributing of the stationary objects and non-stationary objects in the background, the essay has put forward an algorithm of using one or more intensity spaces to donate the stationary objects and non-stationary objects in the background and using the characteristic which the appearance probability of the moving vehicles’ intensity is lower than the background’s intensity, the moving vehicles can be extracted.As to the difficulty which we can not extract the moving vehicles only by using single algorithm in the complicated traffic intersection, first, we can use the algorithm of extracting edge to extract each frame, and then by using the algorithm of sequence coterminous frames differencing we can extract the edge of the moving vehicles. At the same time, the extracted areas of the related edge can be added together, and the interrupting spots which produced by the shaking cameras or coterminous frames differencing can be dispelled, therefore, we can gain the thorough edge information of the moving vehicles. By using the algorithm of coterminous frames differencing, we can gain the outcome image, and then compares it with the moving vehicles’ edge information, at this time, if the gray areas which have background edge can be got rid off and the edge information of the moving vehicles can make up the moving vehicles’ gray areas, the complete extraction of the moving vehicles in the traffic intersection will be achieved.Finally, on the basis of the summarized work, the essay has put forward some useful suggestions in the later study work, which is in great favor of guiding the further study work.更多还原