摘要
LED灯丝外轮廓点的精准特征匹配是实现视觉定位LED灯丝的重要环节。该文为了解决灯丝外轮廓点的坐标定位前期图像处理问题,提出一种基于视觉的LED灯丝特征提取与匹配方法。该方法首先提取LED灯丝清晰的外部轮廓图,然后采用Ransac-Surf角点特征匹配算法完成了匹配。实验表明该算法的立体匹配准确率达到了93.9%,基本满足匹配的实时性要求。该文算法能较好地应用于该类型图像的立体匹配。
Accurate matching of outer contour points is the significant way to realize the visual localization of LED filaments.ln order to solve the coordinate location of the outer contour of the filament,this paper proposed a featrue extraction and stereo matching method for LED filament Based on machine vision.This image processing algorithms extract the cornors feature of the LED filaments,then the Ransac-Surf algorithm can effectively complete the cornor feature matching task.Experimental results show that method presented by this paper can improve the accuracy rate of stereo macthing to reach 93.9%,Basically meet the re- al-time requirements of the stereo matching.This algorithm can be applied to stereo matching of this kind of images.
作者
曹鹏
黄豪杰
CAO Peng, HUANG Hao-jie （Mechanical and Electrical Engineering Department,Foshan University, Foshan 528000, China）
出处
《电脑知识与技术》
2017年第12期190-193,共4页
Computer Knowledge and Technology
