变电金具表面腐蚀膨胀点三维检测算法

Three-dimensional Detection Algorithm for Corrosion and Expansion Points on the Surface of Substation Fittings

为提升配电网变电金具腐蚀膨胀检测的精度,设计了变电金具表面腐蚀膨胀点三维图像检测算法。该算法使用八叉树算法对三维激光图像扫描仪获取的变电金具表面腐蚀膨胀点云图像数据进行压缩预处理,经过法向量计算、局部基面参数化处理、局部曲面拟合等步骤,提取变电金具表面腐蚀膨胀点点云数据的主曲率、形状索引值等局部特征。使用三维霍夫投票算法,对变电金具表面腐蚀膨胀点局部特征实施局部坐标系建立、坐标转换、霍夫投票等处理后,获取腐蚀膨胀点的质心位置,实现变电金具表面腐蚀膨胀点检测。实验结果表明:该算法具备较好的点云数据压缩能力,预处理效果好;变电金具腐蚀膨胀点图像数据存在遮挡缺失时,其特征提取完整度始终高于90%;识别配电站螺丝配件时,特征点匹配度高达100%,识别精度高。

In order to improve the accuracy of corrosion and expansion detection of substation fittings in distribution network,a three-dimensional image detection algorithm for corrosion and expansion points on the surface of substation fittings is designed.The algorithm uses octree algorithm to obtain the image data of corrosion and expansion point cloud on the surface of substation fittings from the three-dimensional laser image scanner.After compression preprocessing,the local features such as principal curvature and shape index value of corrosion and expansion point cloud on the surface of substation fittings are extracted through the steps of normal vector calculation,local base surface parameterization and local surface fitting.Using the three-dimensional Hough voting algorithm,after the local coordinate system establishment,coordinate transformation and Hough voting are carried out for the local characteristics of the corrosion expansion points on the surface of substation fittings,the centroid position of the corrosion expansion points is obtained to realize the detection of the corrosion expansion points on the surface of substation fittings.Experimental results show that the algorithm has good point cloud data compression ability and good preprocessing effect;The integrity of feature extraction is always higher than 90%when there is occlusion loss in the image data of corrosion expansion point of substation fittings;When identifying the screw accessories of distribution station,the matching degree of feature points is as high as 100%,and the identification accuracy is high.