A Traffic Information Video Detection System

Traffic detection system using video analysis, which is an important component of ITS, gains traffic information mainly through the way of image processing. It usually needs several subsystems such as wireless communication system, pan&tilt, GPS and GPRS to cooperate with.This thesis designed and realized the platform of traffic detection system, and researched image processing algorithms related to traffic video detection. Then appropriate algorithm was validated on this platform.A traffic detection system platform was accomplished. Color CCD camera was selected to be the video collection sensor and the TMS320DM642 chip of DSP6000 series to be the CPU. And the video interface, serial, and the EMAC interface were designed. Then SDRAM, Flash and FPGA chips were extended via the EMIF port. The wireless communication subsystem, which took nRF2401 as the communication chip and PIC16F877 as CPU, can realize some functions such as AD, serial, LCD display, and also can collect sensor signals from the road in order to transmit them to the DM642 control board wirelessly. GPS subsystem took ProGin SR-92 as GPS signal receiving module, which can fix position and give time to the system. Then the receiving signal processing can be accomplished by the PIC16F877 of the wireless communication subsystem. GPRS alarm subsystem, which took MC39i as GPRS module, can realize the function of sending alarm messages in certain situations through the operation of hands shaking between DM642 serial. Pan&tilt subsystem, which for realizing closed-loop control of camera pan&tilt, use ADIS16209 to measure angle. Then the results were output to ATmegal6 through SPI interface and then to DM642 through I2C interface in order to determine the movement parameters by analysis and processing with the video.Image processing algorithms related to traffic video detection were researched, such as brightness transformation, image segmentation, feature extraction, and spatial filtering.At last, the various video detection algorithms were tested in the platform. All of that is to provide basic platform and theory, guarantee for developing.