Calibration of full-waveform lidar data by range between sensor and target and its impact for landscape classification

Abstract

Full-waveform LIDAR systems have already been proved to have large potentialities in characterizing the landscape. Especially in the forestry area, more detail information is provided by waveform data processing and new opportunities are inspired for point cloud classification from waveform characteristics. Generally, different objects response to the emitted pulse diversely, which is incarnated in the waveform data. But acquired data is influenced by several factors, so it cannot be directly used in wide area before calibration. Within one flight, some factors such as laser scanner systems, atmosphere conditions, etc. can be considered as constant. Therefore, range between sensor and object could be regarded as one of the most important factor and was introduced to calibrate Gaussian decomposition results of waveform data. Meanwhile, the number of return echoes was also considered in calibration process. After these improvements, the parameters including Gaussian amplitude, standard deviation and energy extracted from waveform data by Gaussian decomposition method were applied for test area classification. A supervised classifier was implemented to distinguish building, grass, conifer and broadleaf. Then the accuracy of the classification results of calibrated and non-calibrated was analyzed, which indicates that the calibrated full-waveform data possibly increase the potential application in landscape identification.