导　　师： 于德新;袁满荣

学科专业： 0823

授予学位： 硕士

作　　者： ;

机构地区： 吉林大学

摘　　要： 随着我国经济快速发展，汽车保有量日益增加，原有的城市道路已经很难满足居民的出行需求，因此交通拥堵现象日益严重，并成为困扰大城市的主要交通问题，对人民生活和经济发展造成了巨大影响。而交通共用信息平台的出现，大大缓解了交通问题，特别是智能交通新技术的不断推广及应用，极大促进了交通拥堵问题的解决。针对以上现象，本文以交通共用信息平台为主要研究方向，针对交通共用信息平台在交通参数提取方法方面，提出了基于视频的车速检测方法和基于遥感图像的道路车辆空间占有率估计方法。 基于视频的车速检测方法是交通共用信息平台的关键技术。由于在实际应用时车速检测系统的实时性及可靠性要求较高，需要视频测速算法具有较好的鲁棒性和较快的运算速度。根据以上条件，本文将基于视频的测速系统划分为视频采集模块、道路运动目标检测模块、Harris车辆角点检测模块、车辆角点匹配模块、坐标变换模块。该方法首先利用混合的高斯模型检测道路运动车辆目标并且提取出车辆的前景图像，再利用Harris角点检测原理检测出前景图像的角点，最后用角点匹配原理对车辆角点进行角点匹配，接着利用单视觉的方法来进行坐标变换，进而得到比较精确的车辆速度。 基于遥感图像的道路车辆空间占有率估计方法是对视频处理技术的补充，也是交通信息获取与处理的关键技术。本文通过对比遥感卫星数据、临近空间数据和航空数据，得出遥感航空数据最容易获取交通信息。因此本文选用遥感航空数据进行仿真，通过对道路图像的增强处理、去噪以及开运算、闭运算、二值化来获取道网结构，利用Hough来估计道路长度，提出利用模糊处理技术来估计车辆长度，最后得到车辆的空间占有率。道路的车道空间占有率是重要的交通信息，要想从遥感图像中提取车道空间占有率就必须先提取道路长度和车辆的长度，所以道路和车辆的长度的提取显得很重要。又由于遥感图像不清晰且有误差，所以本文提出的模糊处理技术是一种较为正确且有效的方法。通过实例分析结果可知，本文构建的两种交通参数提取方法为交通共用信息平台中多源多维信息感知的实现提供了强有力的技术支撑。 With the rapid economic development, the car ownership rates are increasing.However, the urban roads are difficult to expand, and it has caused serious traffic jamsand become the main traffic problem in some big cities, causing a great impact onpeople's life and economic growth. Therefore，the traffic information sharing platformhas been the main idea to solve traffic problems, especially settling the problem oftraffic congestion greatly. This thesis takes the traffic information sharing platform asthe main research direction, from the traffic parameters extracting method, proposingthe detecting algorithm of the road speed based on video and the estimating method ofthe road vehicle space occupancy based on remote sensing image. The detecting algorithm of the road speed based on video is the key technologyfor traffic information sharing system. Due to the reliability and real-time highrequirements of speed detecting system in practical application, video speedmeasurement algorithm has better robustness and faster computing speed. Accordingto the new conditions, this thesis gives processing method in the corner detection,corner matching video speed in the past. In this method, velocity measurement systembased on video is divided into four steps: moving target detection module, videocapture module on the road vehicle, Harris corner detection module, vehicle cornermatching module and coordinate transformation module. The detailed steps of thisapproach for the first use of moving vehicle detection on the road mixed Gauss modeltarget and extract its prospect, and Harris corner detection principle to detect cornersof the foreground image, the corner matching principle angle on vehicle cornermatching, the first rough matching and finally to the coarse matching，then using themethod of single vision to coordinate transformation, and then obtain the moreaccurate vehicle speed. The estimation method of space occupancy of vehicles on roads based on remotesensing image is complementary to video processing technology, and is the key technology of the traffic information acquisition and processing. The thesis draws aconclusion that the aeronautical sensing data is relatively easier to acquire, contrastingwith remote sensing satellite data，close-spatial data and aeronautical data，so the dataof simulation in this thesis takes sample of aeronautical sensing data. To get the roadnetwork structure is by enhancement processing of road image, denoising, openingoperation, closing operation and binarization. This thesis estimates the length of roadsection by Hough and the length of vehicles by fuzzy processing techniques，thusgetting the space occupancy of vehicles. To obtain the space occupancy from remotesensing images, the length of both roads and vehicles must be extracted. Therefore,the extraction of the road length and the vehicle length is very important. Moreover,an effective and correct method about fuzzy processing techniques is introducedbecause of the unclearness and errors in remote sensing image. Through the exampleanalysis results, the proposed two traffic parameter extracting methods provide strongtechnical support for traffic information platform in the implementation ofmulti-source multi-dimensional information perception.

关 键 词： 交通共用信息平台 视频图像处理 车速检测 遥感图像处理 空间占有率

分 类 号： [TP391.41]

领　　域： [自动化与计算机技术] [自动化与计算机技术]