Methods Of New Road Detection Based On FCD And Image Contrast Technology

The new roads are the most important changes, which is one of the main problems of electronic map data updating in location based services, such as car navigation. Currently, the road network is developing rapidly in our country, but the traditional ways of road network data updating meet with the problems of high-cost, long-cycle, less real-time and so on, so it is necessary to improve the efficiency of road network data updating. Hence, the new road detection algorithms are studied in this thesis which based on floating car data.Firstly, the new road detection algorithm based on space grid is studied. Based on the conception that floating car trajectories are restricted to road network, we can get the not matched FCD points after map matching, which can represent suspicious new road. And then, the relationship between not matched FCD points and road network grid can be built up through rectangular grid tessellation, and the grid counter can show the distribution of not matched FCD points in the road network grid, so new roads can be detected after remove some not matched FCD points where the grid counter value is less than setting threshold.Secondly, another new road detection algorithm which based on contrasting images is proposed because above algorithm still have some problems, such as shortage of floating car data and discontinuous between FCD points, and those problems can be solved by generating floating car trajectories. Image of floating car trajectories and road network can be obtained by image contrast technology, and then we can extract image of suspicious new road from above two images by differential treatment, finally, after remove some floating car trajectories which are not long enough we can find out new roads in the road network.Finally, a new road detection platform is built up based on Arcgis9.3, and plenty of new roads are successfully extracted. Meanwhile, some attribute information of new road also can be identified, and the validation result shows that the new road detection algorithm is accuracy and feasibility.