A fast extraction of insulator strings of point cloud of power tower based on multi-view projection
In the automatic inspection of UAV,route planning should be carried out according to the position of in-sulator strings in the dot cloud of power corridor.Due to the large length span of the power corridor and the large number of poles and towers,it is a huge workload to manually mark the position of insulator string.At the same time,the point cloud data is scattered and disordered,and it takes a long time to locate insulator string directly in three-dimensional space.Therefore,this paper proposes a multi-view two-dimensional projection to rapidly extract insulator strings from three-dimensional pole and tower point clouds.Firstly,the robust principal component analy-sis was used to analyze the horizontal direction of the cross arm at the top of the tower in the top view,and the ori-entation of the point cloud of the tower was unified.Then,in the side view,the height of the power line is quickly located by Hoff line detection.According to the spatial relationship between the power line and the cross arm,the corresponding insulator strings are divided into three categories:Top suspension,vertical insulator string and hori-zontal insulator string.Finally,the corner detection is carried out in the multi-view projection,and the connection endpoint of the insulator string with the power line and the tower is located to realize the rapid extraction of the in-sulator string.The method proposed in this paper was used to extract 867 insulator strings from 97 different types of pylons in a 50 km power corridor.It is proved that the method can extract different types of insulator strings quick-ly.
point cloud classificationrobust principal component analysismultivisual projectioncorner detec-tioninsulator string extraction
NETL
NSTL
万方数据
