摘要
针对排水管道视觉检测的研究现状,利用计算机视觉理论,提出一种排水管道侧壁视觉检测三维成像方法.基于排水管道侧壁成像原理,采用Harris角点提取算子进行特征点提取,并利用NCC算法对提取的特征点进行匹配.通过研究基于随机抽样一致性算法(RANSAC)中的预检验快速随机抽样一致性算法(PERANSAC),应用该算法对参数进行预检验,在保证置信度一致的前提下,减少了参与检验的模型参数数量,提高了计算效率.实验结果表明,该方法实现了排水管道侧壁相关图片的特征点提取、匹配和基础矩阵的估计,并得到了三维成像结果.
For the current status of visual detection of drainpipe,a 3D imaging method concerning the visual detection of drainpipe profile was proposed based on computer vision theory.Harris corner extracting operator was used to extract the characteristic points based on the imaging principle of drainpipe profile,and the NCC algorithm was used to match the extracted characteristic points.The parameters evaluation random sampling consensus（PERANSAC） algorithm based on the random sampling consensus algorithm（RANSAC） was discussed.The algorithm was applied to preview the parameters.Thus,the quantity of model parameters in the detection gets reduced and the calculation efficiency gets increased in the condition of guaranteeing the confidence.The experimental results reveal that with the present technology,the extraction and matching of characteristic points as well as the estimating of fundamental matrix for the related image of drainpipe profile are realized,and the 3D imaging results are attained.
出处
《沈阳工业大学学报》
EI
CAS
2010年第2期177-181,共5页
Journal of Shenyang University of Technology
基金
国家自然科学基金仪表专项基金资助项目(60927004)
