Full Waveform Decomposition Of Spaceborne Laser Altimetry To Extract Elevation Control Points In Building Areas

GLAS(Geoscience Laser Altimeter System)is the first spaceborne LiDAR altimeter system for continuous earth observation in the world.It has been widely used in polar glaciers monitoring,land forestry resource surveys,and extraction of elevation control points in flat areas.Due to the large area of the laser footprint,the GLAS original echo signal is the comprehensive superposition effect of the reflection signals of multiple ground objects in the light spot.It is difficult to directly extract the precise elevation value of light spots in the uneven terrain.Therefore,it is very restricted to make elevation control points by using laser altimetry data to assist remote sensing images in non-flat areas,such as building areas.At present,there are few research and application results of using large spot LiDAR altimetry data as elevation control points to assist remote sensing image photogrammetry in non-flat terrain areas,such as building areas.This paper proposes a method of progressive stripping full waveform decomposition and overall fitting to fully decompose the echo waveform,which can accurately characterize the vertical structure of the building area.The high resolution remote sensing image is used to establish the exact correspondence between different high rises of buildings and echo components.Finally,the absolute elevation value of the building feature points is measured with high accuracy by the principles of satellite altimetry and geometric positioning,which is used as the image elevation control point of the building area.The main research contents of this article include:(1)Two key technologies of echo denoising and full waveform decomposition in the processing of spaceborne Lidar altimetry waveform are improved.Aiming at the problems of incomplete or excessive denoising in traditional echo denoising methods,this paper constructs a cascade wavelet transform denoising method with adaptive threshold.Under the same level of wavelet decomposition,the high-frequency noise signal is effectively removed,and the useful signal is not lost in the detail part.Aiming at the difficulty of extracting weak sub-waveforms and the low accuracy of echo feature parameter extraction in traditional full waveform decomposition methods,this paper proposes a method of progressive stripping and overall fitting full-waveform decomposition under the constraints of the total waveform.It can effectively detect the weak sub-waveform,so as to accurately determine the initial value of the Gaussian component,and finally fit and refine the initial value of the component parameter to improve the accuracy of the echo characteristic parameter extraction.(2)A technique and method of Gaussian decomposition wavelet target location based on spaceborne LiDAR altimetry is proposed.For full waveform data,the echo pulse is formed by superimposing the Gaussian components corresponding to a series of ground objects.Obtaining the time difference between the transmitted pulse and the Gaussian components along the echo pulse direction by full waveform decomposition to calculate the accurate distance.The spot position center is geometrically combined with the instantaneous position and attitude of the laser altimeter,and finally the true geographical location of the ground scattering target is obtained.(3)A method of using spaceborne laser altimetry data to assist high resolution remote sensing image to make elevation control points in building areas is studied.Formulating a fine screening rule for high quality laser altimetry points in building areas by using the GLA14 data attribute parameters and GLA01 echo characteristics.Gaussian wavelet components are fully decomposed by progressive stripping full waveform decomposition and overall fitting methods,and the precise correspondence relationship between the building characteristic elevation layers and wavelet components is established by combining high resolution remote sensing images.Finally,according to the laser measurement of the plateau,the relative height of the building area and the absolute elevation value of the equal flat roof layer are calculated,so as to make the elevation control point of the remote sensing images.The paper carried out a more systematic experimental demonstration of the proposed method by using objective indexes,such as root mean square error,signal noise ratio,correlation coefficient,and goodness of fit.The experimental results show that the echo denoising and full waveform decomposition methods proposed in this paper have some advantages over the traditional methods in accuracy,timeliness and reliability.In addition,the median error of the image elevation control points extracted in the building area is 0.6275 m,which can meet the accuracy requirements of elevation control points of the 1:50 000 scale surveying and mapping.Thereby expanding the application range of spaceborne laser altimetry data to assist high resolution remote sensing image photogrammetry.