Research On Key Technology Of Combined Block Adjustment Of ICESat Laser Points And Optical Satellite Imagery

The high resolution satellite series is an important part of the High Resolution Earth Observation System independently developed by China.The high resolution satellite series provides strong support for China's geospatial information construction and basic surveying and mapping.It plays an important role in nation security,disaster monitoring and infrastructure construction.Gao Fen 7(GF-7)will be launched with the highest spatial resolution and positioning precision.It can both achieve three-line array stereo images and laser altimeter data,thus realize linear CCD images collaborative mapping with laser altimeter data.Considering the characteristics of GF-7,it is very necessary and exigent to research on block adjustment of optical satellite imagery assisted by laser altimeter points.With implementation of the national strategy "One Belt and One Road","GlobalMapping" project has been put on the agenda recently.Because of the difficulty of achieving ground control points(GCPs)in China's Qinghai-Tibet Plateau,alpine and gorge regions in Yunnan and Guizhou Province,Northwest China desert areas and the security and territorial rights overseas,block adjustment without GCPs has been a hot-spot technique.In existing research,GIS data(DOM and DEM)and public geo-information data(Google Earth imagery and SRTM DEM)are used to improve the accuracy of block adjustment without GCPs.The NASA Ice,Cloud and land Elevation Satellite(ICESat)is the first laser altimeter satellite designed to monitoring ice sheets changes on the Greenland and Antarctic regions.ICESat both can provide high accuracy laser points and repeatable measurements(the nominal accuracy is 15 cm).Therefore,it is of great significance to use ICESat laser points for block adjustment of optical satellite imagery.At present,block adjustment of optical satellite imagery assisted by laser points is studied mainly on principles and tested in a small range.Besides,there are still some problems in large scale block adjustment without using GCPs,such as limited computation resources,lack of analysis of positioning accuracy,puzzle of image data with abnormal accuracy.Aiming at these problems,key technology of combined block adjustment of ICESat laser points and optical satellite imagery is studied in this dissertation.The main contributions and conclusions include:(1)A multi-criteria constraint algorithm for selecting ICESat laser points as elevation control points is proposed.Firstly,SRTM DEM is using as an elevation reference.GLAS laser points are rejected when height difference between SRTM is greater than 15m and when the ground slope is greater than 10 degree.Secondly,according to the parameters of GLAS product,if i_grav_rcv<100 and i_satCorrFlg is 0 or 1,GLAS laser points are preserved.ICESat/GLAS test data in Shandong Province are used.The test results show:comparing GLAS laser points with 1:10000 scale DEM product,mean of the height difference is 0.296 m,root mean square(RMS)of the height difference is 1.223 m.The automatic measurement method of ICESat laser elevation control points in satellite imagery is proposed.Considering the characteristics of texture of laser spot in satellite images,two measure methods are presented:when the texture is rich,the laser elevation control points are measured with Harris feature extraction and image matching method;when the texture is weak,the laser elevation control points are measured with region growing and image matching method.ZY3-02 image data in Shandong Province are used,and test results show that this method is effective.(2)The large scale block adjustment strategy based on distributed network is studied.Following by ADMM computation frame,large scale adjustment in one cast is divided into different single stereo images processing based on the consistency constraint for parallel computing in network.This adjustment method can make full use of resources in network and can realize parallel computing conveniently.The combined block adjustment model of ICESat laser points and optical satellite imagery is studied.The error observation equations are derived.And the error equations are extended for generalized laser elevation control points.(3)An automatic method based on Bag of Words model for detecting satellite imagery with abnormal precision is proposed.IRS-P5 data,TH-1 imagery and ZY3 imagery are selected to validate this method.Test result shows that all images can accurately registered with the geo-reference images.In this dissertation,ICESat/GLAS data,TH-1 imagery and ZY3-02 imagery in Shandong Province are used to verify the feasibility and validity of the proposed methods.High accuracy GPS points are selected as check points to evaluate the precision of adjustment.The major conclusions are as follows:(1)The elevation precision of block adjustment of optical satellite imagery without GCPs has been significantly improved by adding ICESat laser points as elevation control:without GCPs,the elevation precision of block adjustment of TH-1 imagery is 5.88 m,the elevation precision of block adjustment of ZY3-02 imagery is 3.69 m;adding ICESat laser points,the elevation precision of the former is increased to 2.51 m,the elevation precision of the latter is increased to 1.65 m.(2)GLAS elevation control points along the orbit range of 80 km are enough for block adjustment in flat areas.More GLAS elevation control points should be layout in mountain area.(3)Adding Google Earth imagery and ICESat laser points,the elevation precision on GPS check points of the block adjustment of ZY3-02 imagery:RMS is 1.64 m,mean error is 0.11 m.Adding Google Earth imagery and SRTM DEM,the elevation precision on GPS check points of the block adjustment of ZY3-02 imagery:RMS is 2.32 m,mean error is-1.43 m.Therefore,ICESat laser points are superior to SRTM DEM in elevation control.(4)With the increase of the spatial resolution of satellite imagery,the advantage of ICESat as elevation control is reflected.