Adaptive Unsupervised Classification of TLS Point Cloud Based on Intensity Data

Kai Tan,Xiaojun Cheng
DOI: https://doi.org/10.3788/LOP53.032801
2016-01-01
Laser & Optoelectronics Progress
Abstract:Based on the laser radar range equation and the radiation mechanism of the laser scanner systems,the linear laser intensity correction model is established.Taking into account of the noise impacts and nonlinear effects,a modified correction model is established by revising the parameters,which aims to remove the effects of range and incidence angle to obtain a corrected intensity value associated merely with the target reflection characteristics.By setting the total number of categories,the initial threshold and the threshold step,an adaptive unsupervised classification method is proposed based on the corrected intensity.Experimental results show that the adaptive unsupervised classification method based on the corrected intensity data can accurately classify the point cloud,and the overall accuracy is 84%.
What problem does this paper attempt to address?