The Research Of Vehicle Flow Measuring Based On Video

Nowadays, the increasing problems of traffic speed up the implementation of intelligenttransportation system, in which the traffic information detection system is the very foundation. It isthe variety of traffic flow parameters outputted by traffic information detection system that providea solid base of statistics and analyses for high level intelligent transportation system supportive anddecision-making system, thus reduing traffic flow density of the whole city’s road grid.As a result, it is necessary to design a flexible and effective traffic information detectionsystem to provide accurate and real-time traffic flow information to transportation managementcenter. This thesis firstly compares each kind of current popular detection method and then choosesthe outstanding way of video detection for traffic flow. Next, we also compares some major videodetection systems’ architecture already existed in the market, followed by putting forward a brandnew video diction system block diagram based on Digital Signal Processor (DSP), which willsignificantly reduces the size of detection system as well as enhances flexibility and extensibility ofsystem. And this paper also discusses some difficulties in implemention of hardware schematics.After finishing the phase of hardware design, we explains software development both in DSP andPC, which means that we should develop an embedded operating system on DSP and itscorresponding virtual loop detection algorithm in addition to user interface on PC for usersconvenient interaction with DSP. In order to find best way to run program on DSP, this paper bringsforward a virtual loop detection method based on boundary extraction, and particularly introducesan adaptive edge detection algorithm based on morphology and information entropy. The biggestadvantage of this kind of algorithm lies on its low computation and high accuracy, which exactlysuits for implemention of embedded processor of DSP.