Intelligent Transportation Monitoring Technology Based On Dynamic Image Understanding

The application of dynamic image understanding technology on intelligent transportation is a cutting- edge technology under the domain of present intelligent transportation system, which adds fairly important values both theoretically and practically to this field. Using dynamic visual understanding technology has been proved as a promising trend in the next generation of road traffic detection technology. This paper hereon makes an introduction of omni- directional vision sensor technology, taking advantage of its wide monitoring range, vast information capacity and simple pointing algorithm in order to discover a simple, precise and efficient way to detect road traffic.This paper, firstly introduced the rationale and hardware/software platforms for developing Omni-Directional Vision Sensor. In respect of dynamic image understanding, we set up YUV color space base for mixed Gaussian distribution and background modeling, and then use this model to detect motion targets and to suppress shadow. Next we convert collected images to binary. We then process the binary images based on the theory of mathematical morphology, and calculate out the physical properties of those motion targets. Afterwards, we use such information to location motion range, and design the color characteristic-based tracking algorithm. Our experiments prove that this algorithm takes less calculation workload but has better real time performance, therefore is suitable for precisely detecting transportation information such as traffic, vehicle speed and road procession. With the help of algorithm, we can detect speeding or accidents within the scouting range by applying advanced semantic interpretation on images; hence this research can provide technical foundation for road surveillances and can to some extent help solving merging issues caused by images being shut out.Last but not least, we have developed a testing traffic information collection system based on OpenCV image processing library, Java language, and have presented general features and modules of this system. We have tested this system against real time traffic of west Tianmushan road, Hangzhou and have seen satisfactory results.