Key Techniques Of Data Processing Of Hyperspectral LiDAR For Earth Detection

Full-waveform hyperspectral LiDAR(HSL)collects the spectrum and spatial structure information of targets by actively emitted broad-spectrum laser pulses and detected backscattered signals.With the development of supercontinuum lasers and photoelectric detection technologies,it is technically possible to develop an active hyperspectral lidar system that combines three-dimensional spatial information detection and spectral information acquisition.Compared with the traditional single-wavelength lidar detection technology,which can only obtain distance and geometric information,the new hyperspectral lidar detection technology has shown greater quantitative application potential and adaptability in the fields of vegetation remote sensing.It is becoming more and more popular to study the vertical distribution of vegetation structure and biochemical characteristics.In view of the characteristics and application requirements of the new earth observation hyperspectral lidar system,corresponding hyperspectral lidar data processing is needed to further develop the potential of multi-wavelength backscattered intensity in remote sensing applications.The hyperspectral lidar emits and receives wide-spectrum laser beams that can penetrate the tree canopy through a small gap.This process completely records the waveform data of the impulse response on the laser path,and can provide the distance and spectral distribution of the vegetation.The spatial distribution of more precise and detailed physiological and biochemical information can be revealed from the calibrated waveform data.The preprocessing of hyperspectral lidar data is a key step in detecting target characteristics.In response to the needs of quantitative extraction technology for target feature information,this paper conducts research and analysis on the key technologies of full-waveform hyperspectral lidar data processing.The main work and results of this paper are as follows:(1)Based on the correlation of multi-wavelengths waveform parameter information of hyperspectral lidar,a new hyperspectral waveform decomposition method(HSWD)is proposed,which can extract accurate and complete multi-wavelengths waveform parameters from HSL data.The experimental results show that the HSWD method has better waveform fitting accuracy when the waveform components signal is weak or overlapping.Different from the traditional single-wavelength waveform decomposition method(SWWD),the HSWD method provides complete and reliable waveform parameters through wavelength-dependent hybrid optimization.It can avoids the ambiguity of the parameters between different wavelength channels caused by the missing or wrong detection of some wavelength components.In addition,we also obtained double-echo waveform data sets with different neighboring target distances.The results show that the proposed HSWD method can also maximize the detection ability of hyperspectral lidar systems near targets.(2)By analyzing the phenomenon of pulse signal delay between the spectral channels of targets with different characteristics,a calibration method for pulse delay effect based on the invariance of the red edge channel is proposed.We also analyze the possible sources of pulse delay.The experiment results show that the channel in the red edge spectral region of vegetation has good stability and accuracy for distance measurement for different distances and material properties.Finally,based on the geometric invariance of the specific red-side spectral channel,a practical method for of pulse signal delay effects is proposed.The results show that this method can improve the pulse signal delay effect of full-waveform hyperspectral lidar,while avoiding point cloud ghosting and distance ambiguity between channels.(3)The influence of distance,incident angle and sub-footprint effects on backscattered intensity are analyzed and studied,and an intensity correction method based on laser ratio index is proposed.The performance of the proposed method was tested using the spectral backscattered intensity data obtained by HSL,and compared with the reference target model method.The results show that the laser ratio of the spectral red-side channel to the near-infrared channel has high accuracy and simplicity in eliminating the distance,incident angle and sub-footprint effects,and can significantly improve the difference in spectral backscattered intensities caused by these effects.(4)Through the above research works,we complete the hyperspectral lidar function verification experiment to verify the spatial and spectral information detection of the system.Ranging and spectrum experiments show that the system has high-precision spectrum and structure detection performance.Demonstration experiments show that the HSL system can obtain reflectance in the wavelengths between 540 and 849 nm and have inversion ability of vegetation biochemical components at the two-dimensional leaf level.Furthermore,the construction of three-dimensional chlorophyll and structure has been achieved based on the HSL.The studies shows that a new type hyperspectral lidar system can provide support for accurate three-dimensional vegetation modeling and quantitative remote sensing inversion.