Feature extraction using voxel aggregation of focused discrete lidar data

The ability of multiple-return airborne lidar systems to resolve fine details has grown significantly since their introduction, and many modern instruments are capable of footprint sizes under a half meter. Because most systems scan at near-nadir angles, the pulse origin is often ignored and not recorded with the return point cloud data. By recording this additional position information over multiple focused collects we show how properties such as occlusion can be derived using voxel aggregation of the return data. The voxel map allows us to exploit this new information in scenes with significant spatial structure, such as under tree canopies. Results are presented which show the accuracy of our approach under canopies of varying occlusion levels. Our findings are validated using simulated data provided by the Rochester Institute of Technology through the DIRSIG software.