Machine vision-based lane and vehicle detection technology

Machine vision-based lane and vehicle detection technology
  
The rapid development of the global economy has brought more wealth for all mankind's material and cultural wealth, ownership of motor vehicles is at an unprecedented rate.Automobile as a product of highly developed human civilization, to people's daily lives and also bring great convenience to people's hard to avoid causing incalculable damage to lives and property. Therefore, improving vehicle safety performance, reduce road accidents and improve traffic conditions on the progress of human society and sustainable development one of the important research topics. The main work is the CCD camera device through the car lane on the front of the vehicle real-time monitoring, access to current road traffic scenes video, detection and identification of the vehicle lane and the road in front of goal, vision-based navigation system to provide accurate real-time traffic information . This work includes the following aspects: (1) reviews the research background, pointed out that the detection of lane and vehicle tracking for the importance of reducing accident rates, and a brief description of the current domestic and international road and vehicle based on visual detection and identification system and the development trend of the visual navigation system. (2) For the uncertain reality of traffic scenes, in all weather conditions, real-time detection of lane lines, use color information to improve the lane marking line detection algorithm robustness. And lane by lane marking line in the video images of vehicle shape, size and movement patterns of different vehicles on the lane to eliminate lane marking line in the detection of the resulting interference. (3) in the image sequence, tracking changes in the road, the road taken by the video image sequence information of the consistency of the road from the previous test results rule out a possible follow-up to the error detection frames to improve the accuracy of lane detection results. (4) test vehicles: the full path of the image using the three characteristics of the vehicle - Vehicles bottom of the shadows, there is a pair of vertical edges of the vehicle and vehicle to predict the symmetry of the vehicle and verify the existence of the regional presence of the vehicle. In achieving the vehicle detection based on Kalman filter is detected by the target vehicle tracking. (5) in the Windows platform using Visual C 6.0 development tool, the use of video recording experimental test vehicle, test proven to achieve real-time detection of lane and vehicle tracking. 

Key words: machine vision; intelligent vehicles; lane detection; vehicle detection