Classification of Airborne LiDAR Vegetation Piont Clouds Assisted by Aerial Images

Since it is difficult to automatically distinguish between vegetation and buildings from non-ground point cloud data, this research work proposes a method to automatically classify vegetation in airborne LiDAR (Light Detection and Ranging) point clouds, which is assisted by aerial image. Based on the fact that the spectral characteristics of vegetation are clearly different from other ground objects, digital orthophoto generation and K means clustering algorithm are employed to cluster and enhance the images. Then, the enhanced image and the point cloud data of the corresponding area are fused. Finally, the airborne LiDAR vegetation point cloud data is classified using the image processing results. Experiments are carried out on airborne LiDAR vegetation point cloud data and aerial images of a particular city. Quantitative analysis results prove that total classification accuracy of the proposed method is 96.47%, and the Kappa coefficient is 0.9248. The introduced method can pave the way for automatic classification of the vegetation in LiDAR point clouds.