Elevation control points extraction of spaceborne lasers with multi-feature parameter constraints

The point elevation accuracy of spaceborne laser is the basis of its auxiliary optical stereoscopic image composite surveying and mapping. This study proposes a method for extracting elevation control points of spaceborne lasers based on multi-feature parameters constraints in response to the requirements for the accuracy of laser elevation control points in optical image stereo mapping. This method utilizes the vertical structure information of the target objects contained in the full-waveform data to analyze the characteristics of the high-precision laser elevation control points, and realizes the constraint screening step-by-step based on the validity of data, the number of waveform peaks and echo characteristic parameters. The signal-to-noise ratio, kurtosis and skewness are selected as evaluation indexes. Through the calculation, statistics and analysis of the echo characteristic parameters, the thresholds are determined, and finally the valid waveform data that can be used for spaceborne laser elevation control points can be extracted efficiently. Taking the laser data of GF-7 as an example, such six typical feature samples as grassland, gobi, road, water body, sandy, and cultivated land are selected to determine the suitable thresholds of echo characteristic parameters for GF-7. Taking the airborne LiDAR point cloud data in Jiangsu area as a reference to verify and analyze the extraction accuracy, the experiment results show: based on the multi-feature parameters constraint algorithm, using the set parameter thresholds suitable for the GF-7, the valid waveform can be extracted from the original waveform data efficiently and accurately for the production of high-precision elevation control points. Taking the elevation difference of 0.32 m from the parameter data as the elevation accuracy requirement, the average extraction accuracy is 90.34%, and the average measurement accuracy of the extracted laser elevation is better than 0.5 m.