Geolocation Improvement Of Optical Images Based On SAR Images

With the rapid development of space technology and remote sensing technology,remote sensing has presented multi-sensor,multi-platform and multiangle characteristics.A multi-level,multi-angle,omnidirectional and all-weather global position observation net is forming,which combination of high,medium and low orbits,coordination of large,medium and small satellite,and complementary by coarse and fine resolution.These stereo earth observation network has provided a number of high/medium/ low resolution images.The accuracy of direct position analysis by using satellite imagery is very limited due to system errors such as satellite position deviation,sensor attitude angle observations and distortion,the acquisition of ground control points has the characteristics of high acquisition cost and low efficiency.At present,the research of satellite image positioning at domestic and abroad is still based on homologous image,which not combining the advantages of various satellite images,such as the high plane accuracy consistency of SAR images after calibration.Therefore,it is the main research content of this paper to study how to use the advantages of multi-source satellite data to improve the optical image positioning accuracy after block adjustment of optical image.The main research contents of this thesis are as follows:(1)Based on the imaging principle and geometric model of the CCD sensor,the error sources of satellite remote sensing in the imaging process are elaborated described,and then introduces satellite image geometric processing principle and model,which provided a theoretical foundation for subsequent content.(2)Based on the high plane accuracy consistency of SAR image,the image matching method of heterogeneous remote sensing based on spatial geometric constraints and geometric structure features is proposed for automatic SAR control point extraction.Firstly,a local geometric correction processing is carried out for the optical image by using the rational function model,which aims to achieve local coarse registration between the optical image and the SAR image.Then,a geometric structural feature descriptor is constructed by using orientated gradient information of images to extract the link points,which can be used as ground control point for block adjustment.Finally,four sets of satellite images with different time and resolution are used for the experiment,and the results show that the extracted control points are superior to the two pixels of the image.(3)In the positioning process of block adjustment of optical images,the block adjustment model caused by the absence of ground control points has a high degree of freedom,which leads to the instability of block adjustment accuracy and excessive accumulation of errors.In this paper,the control points that meet the accuracy requirements are introduced for plane control,and the block adjustment positioning model of optical image for SAR image control is established to improve the positioning accuracy of optical remote sensing images.Finally,the block adjustment experiment is carried out with the Gao Fen-2 satellite images locate in Sichuan province,which proves the feasibility and accuracy of the proposed method.