基于FPGA的光条中心线实时提取方法  

Real-time extraction of light centerlines based on FPGA

作　　者：曲岳[1,2] 刘艳滢[1] 王延杰[1] 张雷[1,2] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院大学,北京100049

出　　处：《液晶与显示》2016年第8期818-824,共7页Chinese Journal of Liquid Crystals and Displays

基　　金：国家863高技术研究发展计划(No.2015AA703101013)~~

摘　　要：随着高铁的普及,列车螺钉松动、车身变形等安全问题渐渐引起了人们的重视,如何检测高铁车身是本文要研究的问题。本文采用激光三角法测量模型,当线结构光扫描高铁车身时,可以根据激光条纹的畸变程度,判断高铁列车是否存在安全隐患,而如何快速、精确地提取激光条纹中心线是首先要解决的问题。在考虑了光源选择、环境噪声、被测物反射等因素的基础上,首先对图像进行预处理,选择中值滤波去除噪声,然后用最大类间方差法将目标区域与背景区域分割;最后,用改进的灰度重心法对目标区域进行中心线提取。本文在FPGA上实现了对分辨率为1 024pixel×200pixel图像的处理,实验结果表明,提取光带中心线用时不超过0.97ms,能够满足高铁检测系统对光带中心线提取实时性和准确性的要求。With the popularization of high-speed rail, the screw loosening, body deformation and other safety issues gradually attract people＇s attention, and how to check the rail body is the research field. Using laser 3-dimension measurement model, when the line structure light scans high-speed rail body, it can estimate whether high-speed rail safety hazards exist according to the degree of distortion of the laser stripe. How to quickly and accurately extract the center line of the laser stripe has become the primary problem. Based on the consideration of the factors, such as the selection of the light source, the environmental noise, and the reflection of the measured object, first, we use median filter to remove the noise as pre process, and then target area and background are segmented by OTSU; fi- nally, we use improved gray barycenter method to extract the center line of target area. In this paper the image is 1 024 pixel×200 pixel in FPGA, the experience shows that extraction time of center line is less than 0.97 ms. The algorithm can satisfy the requirements of high-speed rail detection system forlight centerline extraction of real-time as well as high precision.

关 键 词：中值滤波 最大类间方差法 灰度重心法 FPGA 

分 类 号：TN791[电子电信—电路与系统] TP391.9[自动化与计算机技术—计算机应用技术]