The Research Of Lane Line Detection System Base On DM6437

With the social economy and the rapid development of science and technology,people’s income and living standard also improved. In the era of information development, people pay more and more attention to the concept of "time", save time and become the key problem, As a result, people growing requirement for convenient traffic environment, and car ownership will increase greatly, as a direct result of road traffic safety problems become very serious, People pay more and more attention to road traffic safety. With the development of computer vision technology, through the computer vision technology to improve the level of road traffic safety has become one of the important part of the study of countries around the world. Lane line detection system as the basis of intelligent warning system, the system provides an important guarantee for driving safety.In this paper, the hardware development platform to ruitai innovation developed ICETEK DM6437- A4 board as the core, adopting to SONY Super HAD models of image sensor of dahuaDH-CA-F48-E the bolt camera as the input of image acquisition, using image processing technology, through the DSP software development platform for CCS lane line detection and recognition algorithm of writing and research. The main work is as follows:First, this paper introduces the algorithm of TMS320DM6437, CCS software development tools and hardware platform based on hardware platform algorithm optimization method.Secondly, to provide convenient algorithm for lane, we collected for image preprocessing, by comparing the gaussian filtering, median filtering, and the processing effect of median filter, gauss filter is chosen as the image denoising algorithm, this paper to reduce the amount of data, has nothing to do with the goal to choose effective area, in order to enhance image in the driveway, the contrast between the line and background by using the method of histogram equalization is concentrated in the image grey scale of balanced stretching. In edge detection phase,compared with six test operator in a structured way of the edge of the lane line detection in image effect, the paper chose the gauss Laplace operator to detect image edges.Finally, based on the hardware platform using least square method and the traditional Hough transform to extract the lane line in a structured way, when there is interference, the result of the least squares fitting lane line produce larger deviation,while using the traditional Hough transform to extract the lane line effect is better, butthe large amount of calculation algorithm, lead to poor real-time performance, in order to improve its real-time, by capping the polar Angle of lane line search range pole diameter, filter the useless information, reduced the amount of data needed for the calculation. Through the experiment, the improved Hough transform algorithm to improve real-time performance, but still could not reach the requirements of on-board equipment real-time, therefore, put forward based on Gabor transform and initializes the search area detection algorithm, the algorithm using Opencv, realize to the lane line detection in a structured way, through the experiment proves the algorithm of real time in order to achieve the requirements of on-board equipment.