The Study Of Autonomously Auxiliary Speed Control Technology Based On Image Processing

With the development of economics, more and more material requisite especiallyvehicles are required. Along with the constant growth of vehicle possession, the numberof traffic accident is increasing rapidly, which result in great economic losses.Intelligent transportation system for traffic safety is attracting more and more attention.Vehicle assistant driving is an important issue in intelligent transportation system. It is anew interdisciplinary field covering vehicle engineering, automatic control, computerinformation technology, artificial intelligence and so on.Vehicle assistant driving research has made some development at home andabroad, and is being realized for practical and industrial utilization. However, there arestill a lot of technical problems unsolved. Considering both the real-time and practicalrequirements, this paper makes some helpful research on vehicle assistant driving,which can support the later research in the future.This research focus on monocular vision based vehicle assistant driving, whichcan be used in expressway application. Comparing to the costliness and great size ofradar and other sensor technology, monocular vision based vehicle assistant driving is amature technology, and it is cheap and can be installed conveniently. This paper firstlygets the images collected by the camera, then detect and locate the front car byprocessing the images, then realize the distance measurement part through somealgorithm, lastly, uses vehicle crash early warning system to realize warning function.Determine the corresponding relation between images and targets through thecamera calibration; understand internal parameters of the camera for calculation.Introduce some kinds of color models, and choose one suitable model for later use.According to the application environment, firstly determine a security lanes, whichcan greatly reduce the computation time of the image, and then use color model theoryto extract the color region of the car, using morphology to preprocess the licence plateand detect it according to the long proportion threshold. Lastly, roughly determine thepossible location of the licence plate in the next frame.Rectify the images and determine the average height of the character in the licenceplate by calculate the projection of the character in the vertical direction. Then deducean appropriate algorithm and figure out the distance between cars according to cameracalibration theory and distance equation. Research on the vehicle crash early warning system, simplify lane environmentand vehicle condition, propose a better crash early warning algorithm and do somesimulation experiment to achieve better result.