Airborne small-footprint full-waveform LiDAR data for urban land cover classification

Airborne small-footprint full-waveform LiDAR data have a unique ability to characterize the landscape because it contains rich horizontal and vertical information. However, few studies fully explored its role in distinguishing different objects urban area. In this study, we examined efficacy of on land cover classification. The study area is located suburban Beijing, China. Eight classes were included: impervious ground, bare soil, grass, crop, tree, low building, high water. We first decomposed waveform data, from which set features extracted. These related amplitude, echo width, mixed ratio, height, symmetry, distribution. Then, used random forest classifier evaluate importance these conduct Finally, assessed classification accuracy based confusion matrix. Results showed that A was most important feature for classification, other seven features, namely, ? , H Eavg nH R A? SYM S rise ?R f_fl also played roles yielded an overall 94.7%, higher than those previous LiDAR-derived classifications. results indicated could be high-precision proposed help improve accuracy.