Open AccessPole Like Object Detection using PCA inTerrestrial LiDAR System
Author(s) -
Pankhuri Agarwal,
Arshad Husain,
RK Ranjan
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2089/1/012004
Subject(s) - eigenvalues and eigenvectors , ranging , object (grammar) , lidar , principal component analysis , computer science , object detection , process (computing) , artificial intelligence , computer vision , pattern recognition (psychology) , mathematics , remote sensing , geography , physics , telecommunications , quantum mechanics , operating system
Many vertical objects like trees, poles, and pole like structures play a crucial role in the inspection of road safety and planning for road development. The detection of such objects further proves to be helpful in averting roadside accidents and other problems. Light Detection and Ranging technology can be used in identifying these objects. In this paper, we have proposed to detect pole like structures from the dataset generated using a Light Detection and Ranging system. Our proposed pole like object detection approach first segments data into multiple small clusters. The clusters are further analyzed to compute the covariance to identify the linear relationship among the variables. Then eigenvectors and eigenvalues are computed to identify the directions and strength of the data points of clusters. Finally, the Principal Component Analysis approach is used to detect the pole-like structures. The approach is used to identify the target object which uses a threshold value for the angle of the object greater than 70° with respect to the surface. It also uses a normalized eigenvalue equals to 0.7. The efficiency of the proposed is recorded as 93.7%, and the time taken to process the data and detection of the pole-like structures from the dataset is 15 min and 30 sec.