摘要
针对风电叶片,设计了基于图像的裂纹检测系统。采用Harris-SUSAN算法提取叶片图像中的角点,利用SIFT算法确定角点的特征描述子;建立K-D树(K-Dimension Tree),并利用BBF算法(Best Bin First)寻找近邻点和次近邻点,基于欧式距离进行特征粗匹配。然后,采用随机抽样一致算法(RANSAC)进行特征精匹配,剔除错误点对,提高图像拼接质量;用图像的投影特征对横向、纵向和网状裂纹进行分类并计算裂纹参数;通过爬壁机器人搭载工业相机进行系统测试。实验表明,系统可实现裂纹图像拼接和裂纹参数计算。
A crack detection system based on image processing is designed for power wind blades.Harris-SUSAN algorithm is used to extract the corner points in the blade image, and the feature descriptor of the corner is determined by SIFT algorithm. The K-D tree(K-Dimension Tree) is established, moreover the nearest point and the nearest neighbor point are found by BBF algorithm(Best Bin First). The feature rough matching is based on the Euclidean distance, and then the random sampling coincidence algorithm(RANSAC) is used to accurately match the feature. The wrong points are removed to improve the quality of the image stitching. The transverse, longitudinal and mesh cracks are classified by the projection feature of the crack image. The parameters of crack in images are calculated, and the system is tested by the climbing robot equipped with an industrial camera.Experiments show that the system can achieve crack image stitching and crack parameter calculation.
作者
胡世创
魏莹玉
周唯逸
王斌锐
陈迪剑
Hu Shichuang;Wei Yingyu;Zhou Weiyi;Wang Binrui;Chen Dijian(School of Mechanical and Electrical Engineering,China Jiliang University,Hangzhou 310018,China)
出处
《可再生能源》
CAS
北大核心
2018年第8期1231-1237,共7页
Renewable Energy Resources
基金
国家重点研究项目(2017YFC0804609)