Typical Feature Information Extraction of Transmission Corridor Based on UAV Inspection Hyperspectral Image

Abstract

The difference in growth rates of different types of trees around a transmission corridor leads to different levels of threat to transmission lines. In order to develop a reasonable and efficient maintenance plan for targeted needles, a high-spectral aerial survey of 110kv transmission lines was carried out, and high-threat areas with dense trees were selected for analysis. After using the competitive adaptive re-weighting algorithm (CARS) to select the feature bands, the cluster-support vector machine (C-SVM) algorithm is used to classify the four typical trees around the line with the transmission lines, bare soil, and rivers. The results show that the C-SVM classification can get better classification results, and the accuracy of the classification result reaches about 0.95.