Study On The Technique Of Road Detection For Driver Assistance

With the development of society, the number of cars is increasing. The intelligent driver assistance system is an effective method to ensure automobiles being driven safely and reduce traffic accident. In recent years, it has drawn more and more attention form people. And the tecinique of the road detection, which has important significance, is an important research content of the intelligent driver assistance system.Firstly, the principle of camera calibration and the theory of image pre-processing are described in this paper. Secondly, the algorithm of road detection based on texture features is elaborated, which includs the theory of using Gabor filters to extract texture features, the estimation of road vanishing point and the segmentation of road region. In the process of extracting the texture of road images, an improved method which based on four 2-D direction-fixed Gabor filters is utilized. In this way, the complexity of the algorithm is reduced effectively. According to the requirements of practical applications, the design of the driver assistance system based on high-performance FPGA+DSP is presented, and the debugging of the hardware is completed. The main function modules of the system include the video capture module, the core processing module that based on FPGA+DSP, the external memory module and the video output module, etc. The FPGA is the auxiliary digital signal processor, which controls the other function modules and data transmission. The DSP is the core arithmetic processor, which achieves the main algorithm of road detection. Finally, multiple groups of experiments in the view of different road situations are fulfiled and the analysis to the experimental result are given. The experimental results show that the algorithm utilizeed in this paper can accurately complete the road detection no matter in the environment of non-uniform illumination, night or rainy day.