Light detection and ranging (LiDAR) 3-D imaging is used in many fields such as target detection. During the process of echo data collection, large amounts of time of flight (TOF) data are collected (Fig. 1). To save storage space for massive TOF echo data during LiDAR 3-D imaging process, these big data should be compressed using specific compression algorithms while preserving the key target information contained in echo data. Related studies have demonstrated the compression performance under fixed target measurement conditions. However, with technological development, compression algorithms should be improved to meet different application requirements. For this purpose, an improved compression and recovery algorithm based on statistical distribution fitting for LiDAR 3-D imaging echo data is adopted.

The adopted algorithm was designed to process TOF echo data of LiDAR 3-D imaging. The algorithm included two modules: compression and recovery. In the compression module (Fig. 2), the user-provided original TOF echo data and imaging parameters were used as input variables. The polynomial fitting method was employed to calculate the polynomial coefficients of statistical distribution of original TOF echo data. The compression judgment threshold reflected the imaging conditions and was used in the algorithm to automatically separate light counts from dark counts. The compression algorithm stored accurate information about light counts and polynomial coefficients reflecting dark counts, generated byte stream, and compressed it with Lempel-Ziv-Markov chain algorithm (LZMA) for further reduction of file size. In the recovery module (Fig. 3), LZMA was first used to decompress the compressed data to recover the byte stream, which contained user parameters, polynomial coefficients reflecting dark counts, and accurate information about light counts. The algorithm extracted the polynomial coefficients, generated dark counts, and utilized light count information to finally recover LiDAR 3-D imaging data. A miniaturized data collection system (Fig. 4) was designed and developed to collect echo data of LiDAR 3-D imaging to verify the function of the adopted algorithm.

Considering different application requirements, experiments were conducted under different conditions. Under fixed target imaging conditions, the adopted algorithm was verified with different compression judgment thresholds, achieving compression ratios of 1.02%~38.86% and mean square errors between original and recovered data of 0.1238~0.6326 (Table 1). Under indoor target imaging conditions, the adopted algorithm was verified with different compression judgment thresholds, achieving compression ratios of 0.93%~2.06% and mean square errors between original and recovered data of 0.0000~0.4250 (Table 2). The comparison between original and recovered data could be observed from statistical distribution (Fig. 6) and imaging results (Fig. 7) respectively, showing the effectiveness of the adopted algorithm. Therefore, the effectiveness of data compression was demonstrated and verified by the experiments. Compared with previous studies, due to the introduction of polynomial fitting, this algorithm could automatically separate dark counts from light counts without grid division through calibration or prior templates, thereby optimizing the encoding method and improving the algorithm. The miniaturized data collection system simplified the process of LiDAR 3-D imaging, making compression results of individual test cases controllable by users (Table 3).

To meet different application requirements, an improved compression and recovery algorithm based on statistical distribution fitting for LiDAR 3-D imaging echo data is adopted. It is used to process big data collected by the miniaturized data collection system, whose design and development simplify the operation of LiDAR 3-D imaging. By utilizing the polynomial fitting coefficients of the echo data’s statistical distribution and compression judgment thresholds provided by users, the proposed algorithm generates different compression ratios and mean square errors under different application conditions. The algorithm can be used to process big data generated during LiDAR 3-D imaging process and enables users to control the compression results. Future work will be carried out in more complex atmospheric propagation environments to continuously improve the data processing algorithm for LiDAR 3-D imaging echo data.