The Research And Implementation Of In-Vehicle Vision Driving Assistant System

With the development of economic, vehicles has become an important part of people’s daily life. And the security of vehicles has gradually caused wide concern in the society. In order to improve the safety of vehicles, danger warning, security prompt and other active automotive safety technology are getting more and more attention, and the in-vehicle driving assistant system has became a research focus in the active automotive safetv field.The programmable embedded technology has been an important part of electronic information field, and SOPC (system on programmable chip) is one of the most important technologies. The parallel computing and parallel structure of SOPC makes it can process the image very fast and meet the requirement of real-time. With the advantages in image processing SOPC hardware platform provides a good carrier for the in-vehicle driving assistant system.Combining the research of in-vehicle vision and SOPC hardware platform, this paper realized an in-vehicle multi-core driving assistant on SOPC hardware platform, and improved the image recognizing algorithms to make it works better on SOPC hardware platform.The main work in this paper contains three aspects described as follows:Firstly, this paper studied the recognizing algorithms of traffic lights and traffic signs and summarized the advantages and disadvantages of these algorithms. To maximize the advantage of SOPC’s parallelism, we used the color-shape combined algorithm.Secondly, this paper improved the image recognition algorithm to make it works better on SOPC hardware platform. On the basis of color-shape algorithm, this paper realize it with less computation and lower space complexity. In this paper we use Sobel operator and random Hough transfer to recognize the color and the shape of the traffic lights and traffic signs, and this paper also realized the recognition of speed limit signs. Thirdly, this paper realized the in-vehicle driving assistant system on SOPC hardware platform. Corresponding with image recognition algorithm, this paper used multi-core pipelining hardware system to make the system works better. The NiosII soft-core provided by Altera can be well used in implementing the system.