Innovated approach for LiDAR intensity data classification

Airborne Laser Scanning systems with LiDAR technology have been used for acquiring 3D point data of the ground. The capability of LiDAR systems to record the intensity of the backscattered energy added value to the classification of LiDAR data. From pattern recognition research, combined multiple classifiers (CMC) improves the accuracy of the image classification. This paper aims at developing an innovated approach to classify the range and intensity LiDAR data. This approach is a modified CMC where the same classifier is applied on LiDAR data of two data-strips, acquired from different flight lines. The modified CMC is based on the a posteriori probability of the classification results of each data-strip. A study area covering an urban district in British Colombia, Canada, is selected to test the proposed approach. Five different land cover types are distinguished in this area. The proposed approach shows an improvement of the classification results.