摘要
裂缝作为隧道工程衬砌结构中最为普遍的缺陷,具有不断生长延展的特性。以单裂缝为研究主题,采用基于机器视觉的检测系统对同一裂缝不同时期的高效识别可掌握病害发展状况并制定相应的整改措施。因此本文基于裂缝本身特征研究,提出一种裂缝骨架拐点识别的方法。采用改进的八方向Freeman链码技术,在对可疑拐点进行初步识别后,进行伪拐点剔除,得到真实拐点位置。试验证明,针对不同形态的裂缝图像,该算法具有良好的适应性,拐点剔除率达82%~96%,保留了真实有效的拐点。所提取的拐点前后线段长度比值及拐角作为裂缝特征,具有"基因"属性,该特征可用于匹配原始裂缝和延展裂缝,实现对同一裂缝不同时间点的精确定位。
As the most common defects in in tunnel lining,cracks have the characteristics of continuous growth and extension.This paper took single crack as the research subject and used machine vision to identify the same crack in different periods,which can grasp the development of the disease of single crack and formulate corresponding measures.Therefore,based on the study of crack characteristics,this paper also proposed a method of fracture skeleton inflection point identification.By the improved eight-direction Freeman chain code technology,the suspected inflection point was preliminarily identified and the false inflection point was removed.The real inflection point position was obtained.The test shows that this algorithm has good adaptability for fracture images of different types with the rejection rate of inflection point,82%~96%,and the true and effective inflection point is retained.The length ratio of the line segment before and after the inflection point and the inflection corner are extracted as the crack characteristic with gene property.This feature can be used to match the original and extending cracks to realize the precise positioning of the same cracks in different time points.
作者
王睿
漆泰岳
万宇
于海莹
Wang Rui;Qi Taiyue;Wan Yu;Yu Haiying(School of Engineering,Sichuan Normal University,Chengdu 610068,P.R.China;Key Laboratory of Transportation Tunnel Engineering,Ministry of Education,Southwest Jiaotong University,Chengdu 610031,P.R.China)
出处
《地下空间与工程学报》
CSCD
北大核心
2020年第2期524-530,共7页
Chinese Journal of Underground Space and Engineering
