Sparse representation of full waveform lidar data

Full Waveform Data (FWD) has been increasingly becoming available on modern airborne LiDAR systems. Since the waveform signal is noisy and rather sparse by nature, the compressed FWD representation has several advantages. First, the reduced data volume makes the storage and transmission of waveform data faster and more economic. Second, the sparse representation based on proper feature space selection may potentially support the subsequent waveform interpretation and classification processes. Note that discrete return data represent the most basic compressed waveform representation. This study addresses some aspects of FWD compression. First, the wavelet family selection for FWD compression is analyzed, including compression ratio, average/maximum reconstruction errors. Next wavelet filter optimization with respect to typical FWD is investigated. Finally, the performance potential of compressive sampling is assessed along with a brief insight into wavelet representation based waveform classification.