A Research On The Application Of Machine Vision In The Lane Detection Technology

Lane departure warning system, which can warn when the vehicles deviate from the lane by using the machine vision technology, is a part of the safety driving assist system. While lane detection is the key part of the lane departure warning system, this paper researched the machine vision based road images processing techniques, the detection and distinguish of lane boundaries, and the lane departure warning system.We use camera to acquire road images, and use DSP to simulate the algorithms. This paper introduces several common methods of the image enhancement. After experimental comparison, it is obviously that the histogram equalization can effectively improve the contrast of image. Under the situation of noise, the median filter has a better filtering effect and speed. The edge of the filter is clear, and it benefit to subsequent processing. For some classic edge detection algorithms, the descriptions of the principle are given here. Then the paper also gives the comparison results under the different conditions. In order to adapt to the more harsh environment, the edge detection based on wavelet is proposed, which uses the relation of adjacent wavelet coefficients to get rid of the noise and detect the edge. The experiment results prove the advantages of wavelet analysis.On the basis of the preliminary study, the overall procedure of the lane detection is given. We choose the maximum variance threshold method to have a better binary effect. To improve the robustness further, the modified Hough transform is used to extract lane boundaries. Finally, the paper gives a simple introduce for the early-warning model which is used in the lane departure warning system at this stage and design a lane departure warning system with the contact of facts. In this paper, the design of early warning systems get the current position and orientation information of vehicles in the driveway mainly from changes in the angle. According to those parameters, the system decides whether the warning sign is given. Through the simulation results, it’s clear that the method used in this paper is feasible and effectual.