|
LiDAR (light detection and ranging) has recently been recognized as one of the most promising remote sensing techniques due to its excellent performance in the detection of forest inventory, topographic mapping, and automatic driving. Compared to discrete return systems that can provide only range information with a limited number of backscatter intensity value, the full-waveform LiDAR systems record the entire backscattered signals containing more comprehensive geometric physical information, such as depth, shape, and reflectivity, of the target. Waveform decomposition techniques are commonly used to extract attributes of targets from LiDAR waveforms, which consider the received waveform as a mixture of one or more standard probability distribution functions after linearly stretch in the range and magnitude directions. However, the new exponential decomposition algorithm proposed in this paper considers the received waveform as a superposition of several system waveform after linear translation and scaling transformation. The system waveform is fitted by several exponential functions, thus the received waveforms can be treated as similarity transformation of the functions. In the experiment, the method was verified using a system waveform with a negative tail. The significant improvement in range accuracy and reduction of waveform fitting residual indicated that the proposed method can deal with LiDAR waveforms with negative tails and has the potential to extract more accurate structural parameters of the detected object.
|
