立體衛聖映像으로부터 DEM 生成을 위한 嚴密센서모델링 및 REM 技法의 적용 = (The) application of rigorous sensor model and rational function model for DEM generation from stereo satellite lmages

다국어 초록 (Multilingual Abstract)

In this thesis it was applied to base data of sensor modeling with analyzing kinds and quality of orbital characteristic and offered data from IRS-1C PAN(spatial resolution ; 5.8 m), KOMPSAT-1 EOC(6.6 m) and SPOT PAN(10 m) that those spatial resolutions are under 10m.
Being based on this we presented an algorithm to be able to frame economically and efficiently digital elevation model about large area with introducing RFM(rational function model) method applied to rigorous sensor modeling standing on basis of satellite orbit dynamics and collinearity equation, and sensor modeling of high-resolution satellite data like IKONOS.
We were able to improve convergence time and accuracy of the exterior orientation with applying motion characteristic selection stage of the satellite and correlation elimination stage among the exterior orientations sorted from sensor model fit for each satellite data.
In SPOT and KOMPSAT-1 satellite data, the accuracy was highest as average RMSE of check points is ±8.446 m and ±5.615 m in case of modeling with constituting χ(kappa) to 1st order function form and ω(omega), φ(phi) to constant form in perspective center position and satellite attitude parameters, in IRS-IC satellite, it proved that the accuracy of exterior orientation was best, as average RMSE is ±5.730 m in case of modeling with constituting φ, χ constituent to 2nd order function form and ω to constant form in perspective center position.
With introducing bucketing method to improve coefficient determinate confidence and stability of RFM we made a selection of control points, the result to apply constituting model in forms of total 9 kinds from 1st to 3rd to each satellite data at RFM model decision for making DEM was that it was proved that check points and control points are best in case of constituting RFM model formed that SPOT image is ρ2≠ρ4 or ρ6≠ρ8 in 2nd order function, KOMP5AT-1 image and IRS-1C image is ρ2≠ρ4 or ρ6≠ρ8 in 3rd order function.
DEM was generated with kriging interpolation extracted three dimensional ground coordinate to rational quadratic function form, we compared it to reference digital elevation model made from 1:5, 000 digital map, and so, could generate digital elevation model in the accuracy as average RMSE of elevation was ±9.76 m(SPOT), ±11.34 m(KOMPSAT-1), ±11.61 m(IRS-1C) in rigorous sensor and ±13.69 m(SPOT), ±13.03 m(KOMPSAT-1), ±14.65 m(IRS-1C) in RFM.

목차 (Table of Contents)

• 목차
• Table Contents = v
• Figure Contents = ix
• Notations = xiii
• Abstract = xix
• I. 서론 = 1
• 1. 연구배경 및 필요성 = 1
• 2. 연구의 목적 = 4
• 3. 연구동향 = 8
• 4. 연구의 범위와 방법 = 14
• II. 위성시스템의 특성과 3차원 지상좌표 추출 이론 = 17
• 1. 위성시스템 및 데이터구조 = 17
• 1) IRS-1C위성 = 17
• 2) KOMPSAT-1 위성 = 26
• 3) SPOT 위성 = 33
• 2. 3차원 지형정보의 좌표계간 상호변환 = 39
• 1) 기준좌표계의 상호변환 = 39
• 2) 좌표계간 상호변환의 적용 및 평가 = 51
• 3. 입체위성데이터로부터의 3차원 지상좌표 추출 이론 = 56
• 1) 시차공식 = 56
• 2) 위성 센서모델링 = 58
• III. 적용기법 = 67
• 1. 엄밀센서모델링 = 69
• 1) 위성 궤도요소의 모델링 = 69
• 2) 내부표정 = 78
• 3) 외부표정 = 82
• 4) 위성 센서모델링 = 88
• 2. RFM(Rational Function Model) = 97
• 1) Rational Function = 97
• 2) 최소제곱법에 의한 반복해 = 100
• 3) Rational Function의 공간교차이론 = 103
• 3. 자동매칭기법 = 108
• 1) 표준상관 매칭기법 = 109
• 2) 최소제곱 매칭기법 = 111
• 3) 영상피라미드 매칭기법 = 114
• IV. 실험 및 결과분석 = 117
• 1. 사용데이터 및 관측점의 선정 = 117
• 1) 사용데이터 = 117
• 2) 센서모델링을 위한 관측점의 선정 = 124
• 3) 연구대상영역의 선정 = 134
• 2. 자동매칭기법의 적용 = 136
• 3. 엄밀센서모델링에 의한 DEM생성 = 142
• 1) 위성궤도요소의 산출 = 144
• 2) 위성센서모델의 결정과 평가 = 151
• 3) DEM의 생성 = 158
• 4. Rational Function Model에 의한 DEM생성 = 163
• 1) Rational Function 계수 산출 및 모델 선정 = 164
• 2) RFM의 결정 및 평가 = 167
• 3) DEM의 생성 = 169
• 5. 수치정사영상의 작성 = 173
• 6. DEM의 평가 = 175
• V. 결론 = 187
• 참고문헌 = 189