摘要
针对车道线识别算法复杂、计算量大、软件处理慢等问题,将FPGA并行处理技术与数字图像处理技术相结合,完成车道线识别系统设计.整个系统分为中值滤波、二值化、骨架化、二次滤波拟合四大模块.创新性提出了"双对分最大类间方差滤波迭代算法"实现图像二值化阈值选取;另提出"基于行逼近的最小二乘曲线拟合算法"以完成次近景的车道线提取;并对直线识别算法进行改进以完成近景的车道线提取.仿真结果表明本算法识别准确,符合实际路况,且可满足系统实时性要求.
According to the complexity of lane identification algorithm, large amount of calculation and slowness of software processing, the paper combines FPGA parallel processing technology with digital image processing tech- nology, completes the lane line identification system design. The whole system is divided into four modules,which are median filtering, binarization, skeletonization, secondary filter and fitting. ＂Double split iterative algorithm of OTSU＂ is innovativly proposed to realize the algorithm of design of parallel threshold value selection. And ＂fitting method of the curve line of least squares approximation based on line approximation＂ is put forward in the paper to complete closer shot lane recognition. Besides, we improve the algorithm of straight line recognition in close shot. Simulation results show that the algorithm to identify accurate, in line with the actual road conditions, and can meet the real-time demand in system.
出处
《哈尔滨理工大学学报》
CAS
2013年第6期74-79,共6页
Journal of Harbin University of Science and Technology
基金
黑龙江省教育厅科学技术研究项目(12531107)
关键词
FPGA
车道线识别
并行阈值选取算法
曲线拟合算法
FPGA
lane line identification
algorithm of parallel threshold value selection
fitting method ofthe curve line of least squares approximation based on line approximation
