| Over the past decade, China has accelerated the process of urbanization, the construction of the road network as the main form of large cities and urban agglomerations has made considerable progress. However, traffic congestion has become a chronic illness of city, traffic accidents are common. Strengthening the monitoring of the road networks in urban agglomerations, and researching the traffic bottlenecks which have heavy traffic congestion queue for characterization, can provide reliable data for planning, counseling and optimization of road network in traffic management sector, and can be basic work to solve these problems.But urban agglomeration road network is number of points, lines long, wide, only rely on the coil, video, and other traditional means of monitoring are difficult to fully grasp a wide range of monitoring in the economy, macroeconomic and other aspects. Although there have many traffic monitoring devices in some cities of urban agglomerations, and have high degree of traffic monitoring information, relying solely on traditional traffic monitoring tool, it is difficult to achieve unified management in a short term. Meanwhile, the long-term management of administrative compartmentalization between adjoining cities, urban fringe road network often become the monitoring blind spot, traffic data is difficult to share, video and other data is difficult to meet the needs of long-term preservation.The ZY-3 was successfully launched in January 9, 2012 as the representative of the high resolution remote sensing technology had achieved major breakthrough in our country. Extracting traffic information from high resolution remote sensing images, and applying to a large area of urban agglomerations traffic road network management, has become a research hotspot. In this regard, my research which has the support of the Ministry of Transport of the project, has researched technology of the road network bottleneck identification in urban agglomerations, the main contents are as follows,This paper studied the status of the current transportation applications of remote sensing technology in domestic and international, cleared the purpose and meaning of the research, then developed research and technology roadmap.I studied the common identification norm of the traffic bottleneck which always has heavy traffic congestion vehicle queue in the urban agglomeration. Based on the characteristics of the traffic information which detected from the high resolution remote sensing images, established a road traffic bottleneck identification technology which has an indicator, the indicator is vehicle queue length ratio.After the vehicle information extraction, interrupt the road network to establish vehicle search block road, and let the vehicle congestion queue occupancy 0.3 as a threshold value to traversal search vehicle queue in a large area image, thereby we formed a slider threshold discriminant of the vehicle queue detection in high resolution remote sensing images. We based on a long time, and more time data, according to the frequency of vehicle queue length ratio occurrence, as Shenzhen an example, to build traffic network bottlenecks recognition technology.For the extraction of vehicle queue which on the basis of the detection of vehicle from the high resolution remote sensing images, we developed two kind of vehicle detection method were based on aeronautics and aerospace high resolution remote sensing images. Based on the differences between light and dark brightness feature in aerospace images, this paper developed an automatic dual-threshold extraction algorithm. And we developed a method combined SURF(Speed-up robust feature) and SVM(Support vector machine) to extract vehicle target from aeronautics images. This method uses the information-rich features in aeronautics images, and constructs 400 positive and 400 negative samples. And this method needs to extraction those samples’ SURF, uses SVM to train high-dimensional mapping model, then extract the vehicle target from traversal search in the interested area which has marked uses road network vector in aeronautics images. Contrast visual interpretation, two vehicle target extraction methods’ false detection rates are 10.2%, 16%, missing rate are 5.1%, 15%.I developed a system which can be applied to judgment and research of the road network bottlenecks in urban agglomerations. The system consists of three subsystems uses C++ programming language, Arc GIS Engine for basic data processing platform. The subsystem contains information extraction module, information display module, and the detection of road bottlenecks module. Key Words: Urban agglomeration,Road bottleneck detection, High resolution... |
PDF Full Text Request