Detection Of Ahead Vehicle Obstacle Based On Radar And Computer Vision

Intelligent vehicle is becoming the focus of social studies. Safety driver assistance technology is an important part of intelligent vehicle. Safety driver assistance technology can timely and accuratly provide warning to the driver.Meanwhile the driver can get the speed status information. Safety driver assistance technology can enhance the vehicle safety efficiently.The dissertation establishes the fusion of detection of ahead vehicle obstacle based on radar and computer vision.Firstly, the millimeter-wave radar obtains the obstacle information from the primary principle that the same lane information. Using the life cycle approach to achieve accurate selection of the valid target.Secondly, through a combination based on Haar-like rectangle features and Adaboost classifier machine vision recognition algorithm to complete the identification of the vehicle in the region of interest created.Meanwhile based on the target features and Kalman filter to track the valid target for real-time.Again, to build a model of integration based on millimeter-wave radar and machine vision for the integration of data in space and time.Finally, be a real vehicle test to vertificate the fusion model.The experiments are going in a variety of different weather, lighting and road conditions.The experimental results indicate the fusion model in this dissertation can be real-time and effective.