Research On Uncontrolled Positioning Of High Spatial Resolution Remote Sensing Image

The accuracy of remote sensing image ground target positioning depends on the number of control points and their optimized spatial distribution.And the data of ground control points is not easy and difficult to obtain,such as abroad areas,islands(reefs)and deserts.Therefore,how to improve the positioning accuracy of remote sensing image(satellite image,uav image)without ground control points-that is the remote sensing image of high spatial resolution without GCP has become an important research content.Based on the analysis of the results of the remote sensing image of high spatial resolution without GCP,this paper designed two schemes for the positioning of the remote sensing image without ground control points for different data types.The main work and innovation contents are as follows:1.image non-controlled positioning technology based on POS data.In the method,coordinate system transformation,POS data error analysis,error correction model and other aspects are discussed and studied in depth:(1)In this paper,the factors affecting the positioning accuracy of uav image are studied.The experiment proves that the direct image positioning based on POS data in the absence of control points has major system error.(2).Through the analysis of the error source of POS data,determining the system error of POS data and the element error of the external alignment line of the image between odd and even bands has the opposite in the process of flight,error correction model is put forward and designed a kind of error correction scheme.System error compensation model for calculating according to the parameters,the model coordinates obtained from the forward intersection of uav image are compensated for errors.The experiment proves the effectiveness of the scheme.2.The technique of remote sensing image positioning assisted by DEM data.According to the process of this scheme,important research has been done on DEM evaluation,DEM matching algorithm,RPC error model,etc.The main contents and innovations are as follows:(1)The positioning principle of traditional remote sensing image is systematically studied.The rational polynomial model(RPC)and error compensation scheme are expounded.It is concluded that if only RPC model is used for image forward intersection without control,the image positioning accuracy cannot meet the practical requirements.Put forward based on DEM has the stable control information,the image can be corrected without control points.(2)A method of DEM matching without control based on symbiosis matrix is designed.By converting DEM into grayscale image for matching,this method can enlarge the data matching range without setting the initial value of iteration,effectively improve the iterative efficiency of the algorithm and DEM matching accuracy,and verify the matching results in frequency domain with phase correlation,and the accuracy of this method is proved by experiments.(3)Put forward a kind of remote sensing image orientation assisted DEM data。Firstly,the DEM/DSM extracted from the stereo image pair was matched with GDEM or other existing DEM data in the experimental area Secondly,using the conjugate points in the reference DEM as the control data,the RPC compensation model was established.Finally,the model parameters are calculated according to the conjugate points,and the system error compensation is carried out to improve the positioning accuracy of satellite image without ground control points.The experiment proves the feasibility of the scheme.