A Study On The Positioning Theory Of High-Resolution Satellite Imagery Based On Linear Features

The paper makes a comprehensive investigation on the high precise positioning of High Resolution(HR) satellite images, and the emphasis is on how to improve the positioning precision of their rigorous geometric and RPC models with the replacement of straight control lines to normal ground control points, as well as on the on-orbit geometric calibration of HR remote sensing satellites. The main work and innovations are as follows:1. The imaging mechanism and characteristic of HR satellite CCD sensors is analyzed at first in detail, then the fundamental theories to build up the rigorous geometric model of HR satellite image is discussed, and ZY-3 panchromatic stereo images are used to validate its correctness and precision.2. The basic principles of line feature photogrammetry is investigated. According to the varying characteristics of satellite CCD sensor exterior orientation parameters, an extended mathematic model which is geometric constrained by straight line features is established for the exterior orientation and on-orbit geometric calibration of HR satellite images. Experimental results show that this model is of equivalent geometric accuracy with traditional model.3、A systematic error compensation scheme of RPC model based on straight control lines are put forward and tested using actual data. Test results indicate that the compensation effects of the proposed method is the same as classic method based on ground control points. Thus provides another precision optimizing scheme for the RPC model of HR satellite images.4. With the combining of the error model of satellite CCD sensor interior orientation parameters to the positioning model based on straight line feature proposed above, another self-calibration adjustment system for the on-orbit geometric calibration of HR remote sensing satellite is put forward, which can improve the positioning precision of satellite image in advance.