一种照度自适应车道检测算法及多核平台实现  

An Illustration Self-adaptive Lane Detection Algorithm and Multicore Platform

作　　者：谢红韬 杨斌[1] 张翠芳[1] 

机构地区：[1]西南交通大学智能系统与控制实验室

出　　处：《单片机与嵌入式系统应用》2017年第7期23-27,共5页Microcontrollers & Embedded Systems

摘　　要：针对车道线检测，基于图像白平衡算法和灰度直方图，自适应地提取出感兴趣区域，并自适应确定Canny边缘检测算法的高低阈值。通过对概率霍夫变换得到的直线点集进行RANSAC拟合，满足了在不同光照条件下的自适应车道线检测，并基于英伟达Jetson TK1嵌入式开发板结合开源GUI库Qt，使用其Qt Quick开发出一套车道线检测系统。Aiming at the problem of lane detection, the algorithm is proposed which can adaptively extract the region of interest and deter- mine the high and low threshold of Canny edge detection algorithm, the algorithm is based on the image white balance algorithm and gray histogram.Through the RANSAC fitting of the linear point set obtained by the probabilistic Hough transform, the algorithm can meet the requirements of the adaptive lane detection under different illumination conditions.In the paper, the Qt Quick is used to develop a lane detection system based on the NVIDIA Jetson TKI embedded development board and the open-source GUI library Qt.

关 键 词：车道线检测 CANNY边缘检测 霍夫变换 Jetson TK1 QT Quick 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]