Study On Transmission Line Corridor Change Detection Based On Airborne LiDAR Data

Regular transmission lines inspection is a key issue to ensure its safe operation for the power grid operation and maintenance management department,and while the 3D information acquiring and analyzing plays an important role during high accurate inspection procedure.Li DAR has been widely used to meet the demands of geospatial 3D information of transmission lines due to its unique capability of obtaining high density and accuracy of 3D data directly and quickly.Meanwhile,it can penetrate vegetation canopy to detect terrain information of sub-canopy area.Nowadays,remarkable results have been achieved in many related aspects,such as transmission line elements classification,safe distance detection and 3D reconstruction.However,Li DAR data application on transmission lines is mostly limited to classification,modeling and static analysis tasks of one dataset,which ignores the association between multi-temporal data and deep mining of available information,resulting in data waste and redundancy.This thesis focuses on the key technologies of transmission line corridor change detection,provides more support for transmission line corridor change detection and 3D visual security management system construction.The main contents are as follows:(1)Pre-processing technology for airborne Li DAR data of transmission lines.In view of the characteristics of large topographic fluctuation,dense vegetation and irregular distribution of point cloud,the paper proposes a denoising algorithm based on noise classification.Next,it focuses on the improved progressive TIN densification filtering method for complex environmental features,such as large terrain fluctuations.Compared with traditional method,the proposed algorithm performs well and experiments show that the errors of type I,type II and total are reduced by 5.43%,0.89%,3.19% respectively.(2)Feature recognition of transmission line based on random forest classification method.After pre-processing of airborne Li DAR data,this paper utilizes a random forest classification method based on the integrated thought to divide the transmission line features into three categories: vegetation,power lines and power towers.Experiments show that the classification method has achieved a good result with an accuracy over 98%.The technique flow provides support for the visualization of change detection results.(3)Automatic point cloud registration method from coarse to fine based on the transmission line features.Accurate registration of different times of Li DAR data plays a key role in 3D change detection.The principal component analysis method is used to perform the main axis analysis of the tower to realize the rough matching of tower point cloud.Next,a refined iterative closest point method is used to complete the precise matching.Finally,the unification of the coordinate system is realized with the transformation matrix.(4)Visualization of the 3D change detection and experiments based on the registered data.The results show that the method performs well with the advantage and potential of high efficiency and high accuracy,and has potential value for transmission line corridor change detection.