i
ii
iii
iv
v
vi
vii
1
2
3
4
Image Processing Image Pyramid Edge Detection Epipolar Image Image Matching LIDAR + Photo Cross correlation Least Squares Epipolar Line Matching Low Level High Level Space Intersection Virtual LIDAR Data Building Extraction 5
6
Ranging Unit Scanner Control, Monitoring and Recording Units LASER FOOTPRINT DGPS IMU flight directi on swath width scan direction 7
A T A T t t A R A R t t t L t L (a) (b) A Linst h θ inst dx across 8
dx along 비행방향비행방향비행방향 9
f L y x a o a x y a Z L Z Y A X A Z A X L Y A Y L X 10
f (X L, Y L, Z L ) (X, Y, Z ) (x, y ) (x 0, y 0 ) r ij R i j R 11
P S P L Q P R L 왼쪽사진투영중심 f Q L E L Epipolar Axis E R Q R f R 오른쪽사진투영중심 S 12
O O' P P' A P P' y y y 13
(r, c ) (x, y ) R N = R B R T R R T R N R N 14
R N 15
f P = f N Z Z 수치영상 y P 수치영상 y P p P p P x P x P f P Y f P Y X X -f N y N 수직영상 -f N y N 수직영상 x N x N p N p N y 16
x y 표정작업전 항공사진 표정작업후 normalized 사진 projective 좌표변환 ( 기하학적좌표변환 ) Affine 좌표변환 에피폴라영상의크기설정 수치영상 normalized 영상 영상재배열 (resampling) (a 1, a 2 ) A A 17
L1 L2 a1 a2 Z o1 o2 Y XA A ZA YA X 18
Image Profile of a horizontal line First derivative Second derivative (a) (b) 19
20
ρ ρ ρ ρ = σ LR σ L σ R σ LR L R σ L L σ R R ρ 1 ρ 1 ρ 기준영역 Template column 정합대상영역 Search area column initial point matching start point target point row row 21
g L ( x, y) g R ( x, y) g L g R σ L σ R σ LR σ L = σ R = g L = Σn i =1Σ m j =1g L ( x i, y j ) n m g R = Σn i =1Σ m j =1g R ( x i, y j ) n m Σ n i =1Σ m j =1 g L ( ) n m 1 Σ n i =1Σ m j =1 g R ( ) n m 1 x i, y j g L 2 x i, y j g R 2 σ LR = Σn i =1Σ m j =1 g L ( x i, y j ) g L g R ( x i, y j ) g R n m 1 n m 22
(x s, y s ) g t g s 23
기준영역 Template column 정합대상영역 Search area column initial point target point matching start point row row h 0 h 1 g t (x t, y t )= h 0 + h 1 g s (x s, y s ) + n (x s, y s ) g t (x t, y t ) (x t, y t ) g s (x s, y s ) (x s, y s ) n (x s, y s ) h 0 h 1 X = a 0 + a 1 x s + a 2 y s Y = b 0 + b 1 x s + b 2 y s 24
a 0 a 1 a 2 b 0 b 1 b 2 g t = h 0 + h 1 g s [( a 0 + a 1 x s + a 2 y s ), ( b 0 + b 1 x s + b 2 y s )] + n(x z,y s ) g t g s x s y s F = h 0 + h 1 g s [( a 0 + a 1 x s + a 2 y s ), ( b 0 + b 1 x s + b 2 y s )] F F = F 0 + F h 0 + F h 1 + F a 0 + F a 1 + F a 2 + F b 0 + F b 1 + F b 2 = F 0 + dh 0 + g s (X, Y ) 0 dh 1 + h 1 g s (X, Y ) X + h 1 g s(x, Y ) Y (1 + x + y) 0 (1 + x + y) 0 g t (x t, y t ) g s (X, Y )=dh 0 + g s (X, Y ) 0 dh 1 + h 1 G X0 + h 1 (G x x ) 0 + h 1 (G X y) 0 + h 1 G Y0 + h 1 (G Y x ) 0 + h 1 (G Y y) 0 h 1 G X0 = da 0 h 1 (G X x ) 0 = da 1 h 1 (G X y ) 0 = da 2 h 1 G Y0 = db 0 h 1 (G Y x ) 0 = db 1 h 1 (G Y y ) 0 = db 2 25
AX = L + V A = 1 g s (X,Y) 1 h 1 G 1 X 0 h 1 G X x 1 0 h 1 G X y 1 0 G 1 Y 0 G Y x 1 0 G Y y 1 0 1 g s (X,Y) 2 h 1 G 2 X 0 h 1 G X x 2 0 h 1 G X y 2 G 2 0 Y 0 G Y x 2 0 G Y y 2 0 1 g s (X,Y) n h 1 G n X 0 h 1 G X x n 0 h 1 G X y n 0 G n Y 0 G Y x n 0 G Y y n 0 X = dh 0 dh 1 da 0 da 1 da 2 db 0 db 1 db 2 L = g t (x t, y t ) 1 g s (X, Y) 1 g t (x t, y t ) 2 g s (X, Y) 2 g t (x t, y t ) n g s (X, Y) n 3 3 AX = L + V X = ( A T A) 1 AL σ 2 V = T V n m 8 26
(x, y ) 27
28
29
180 y 30
2 4 5 4 2 4 9 12 9 4 5 12 15 12 5 4 9 12 9 4 2 4 5 4 2-1 -1-1 -1 8-1 -1-1 -1 31
(a) 원영상 (b) canny filter 적용결과 (c) threshold 100 으로이진화 (d) thinning 적용최종영상 32
(a) 원영상 (b) laplacian filter 적용결과 (c) threshold 100 으로이진화 (d) thinning 적용최종영상 33
34
35
6 1 5 2 pixel 4 7 3 8 original matching point 1 2 3 4 1 2 3 4 epipolar line matching points edge 3 14 3 14 16 2 13 16 2 13 15 1 12 15 1 12 6 5 4 10 11 6 5 4 10 11 17 7 20 17 7 20 18 8 21 18 8 21 19 9 22 19 9 22 36
37
(a) original LIDAR data (b) virtual LIDAR data 38
(a) original LIDAR data (b) virtual LIDAR data 39
(a) Left Photo (C5-11) (b) Right Photo (C5-10) 40
41
42
43
(a) LIDAR 데이터 (b) 건물추출결과 44
(a) 가상 LIDAR 데이터 (b) 건물추출결과 45
(a) LIDAR 데이터 (b) 건물추출결과 (a) 가상 LIDAR 데이터 (b) 건물추출결과 46
47
(a) LIDAR 데이터 (b) 건물추출결과 (a) 가상 LIDAR 데이터 (b) 건물추출결과 48
49
(a) building A (b) building B (c) building C (d) building D (e) building E (f) building F 50
51
52