Research Of Peccancy Vehicles Capture Algorithm In Mobile Intelligent Transport System

With the constant improvement of the level of national economy, the number ofprivate cars is increasing in the city. The phenomenon that private cars illegally takeup bus lanes has occurred, has affected public transport unobstructed sex and safety.In order to solve this problem, the traditional way of fixed point monitoring isdifficult to finish the work of violations monitoring and forensics due to its ownlimitations, while the mobile intelligent bus system provides a good solution to aboveproblemsThe mobile intelligent bus system consists of the front-end capture subsystem andback-end data processing subsystem. The front-end capture subsystem mainly beresponsible for capturing peccancy vehicles and converying the capturedinformation.Thr front-end capture subsystem uses the TI company’s DSP DaVinciDM6467as control board. Lane line detection algorithms and peccancy vehiclesdetection algorithm was tested and accomplished in the platform. So the researchcontents and innovations are as follows:In this paper, the design of peccancy vehicle detection algorithm is one of theapplications for mobile intelligent transportation system. The front-end capturesubsystem is located in the bus processed the image of ahead road, so the lane linedetection and vehicle detection algorithms are completed in a dynamic and complexenvironment(1) In this chapter, we have introduced the hardware structure of algorithmtransplantation platform DM6467, and completed the construction of the algorithmdevelopment environment. On the platform we carry out a experiment above thepeccancy vehicles capture algorithms, and completed the research and implementationof the algorithm(2) Lane line detection. In this paper, we have inproved the traditional Houghtransform combined with the lane line of prior knowledge on the basis of the traditional Hough transform, reducing the computation burden of the Houghtransform,What’s more we apply it to the lane line detection algorithm, and make thealgorithm more conform to the requirements of the DM6467platform transplantation.The simple and practical detection method of lane model is adopted, to test the lineafter the improved Hough transform, and achieve the lane line detection algorithm onthe DSP platform the research and implementation(3) Vehicle detection. In view of the requirement of real-time and the limitationof the DaVinci platform processing power and storage space. In this paper, a vehicledetection method based on feature area is raised. Firstly transform video images fromthe RGB color space to HSV color space, and extract of H tonal from componentdiagram, the H color components of red in binarization. The department is thecharacteristics of the vehicle area. Using the sweep algorithm based on scan line toextract characteristics of the vehicle are. By the red car taillight filter and matchingalgorithm realize vehicle detection. Vehicle detection algorithm based on feature areais smaller amount of calculation, and through the DM6467platform test, the accuracyrate of vehicle detection is above90%(4) Vehicle lock to weight algorithm. After the detection of peccancy vehicles, ifdirectly capturing illegal vehicles, repeat capture phenomenon will appear. So weneed to deal with the weight of image. This paper proposes a vehicle lock to heavyalgorithm based on feature area, and realize on the DM6467platform...