LiDAR A utomatic D etection for Misclassified A erial LiD A R D TD

저작자표시 - 비영리 - 변경금지 2.0 대한민국 이용자는아래의조건을따르는경우에한하여자유롭게 이저작물을복제, 배포, 전송, 전시, 공연및방송할수있습니다. 다음과같은조건을따라야합니다 : 저작자표시. 귀하는원저작자를표시하여야합니다. 비영리. 귀하는이저작물을영리목적으로이용할수없습니다. 변경금지. 귀하는이저작물을개작, 변형또는가공할수없습니다. 귀하는, 이저작물의재이용이나배포의경우, 이저작물에적용된이용허락조건을명확하게나타내어야합니다. 저작권자로부터별도의허가를받으면이러한조건들은적용되지않습니다. 저작권법에따른이용자의권리는위의내용에의하여영향을받지않습니다. 이것은이용허락규약 (Legal Code) 을이해하기쉽게요약한것입니다. Disclaimer

LiDAR A utomatic D etection for Misclassified A erial LiD A R D TD

LiDAR A utomatic D etection for Misclassified A erial LiD A R D TD

(Digital Elevation Model), 3.. LiDAR(Light Detection And Ranging) (Automatic Filter),,.,. LiDAR. (Digital Terrain Data),. (Commission Error) (Omission Error).,. LiDAR (Voxel Structure), Eigen Planarity, Voxel Density, Voxel DeltaZ, WENS Z 4.., LiDAR,. LiDAR.

Abstract Airborne laser scanning technology has been used in various applications. For example, Digital elevation model, Contours and 3D urban modeling. Among them, fast and accurate extraction of ground points is considered as the most important issue for reconstruction bare-earth. To do this, a lot of automatic filters are studied to extract the ground points from Airborne LiDAR(Light Detection And Ranging) data. This is extracted the ground points based on the assumption that the ground is a relatively flat and using parameters such as the height distances and angles between the points. However, the accurate extraction still has limited due to the complexity of the real world and irregular geometry of LiDAR data. To overcome these problems, we required costly manual processing or using aerial images. It should lead to high costs for LiDAR data post-processing works. In this study, the efficiently method is proposed to compliment the limitations of automated filters and to check the accuracy of the LiDAR data filtering. The method automatically detect the misclassified candidate areas. Typically, the filtering errors divided into commission error and omission error. These errors must be and corrected, it is required the exact filtering performance of LiDAR DTD. The Voxel Structure was applied and the some parameters were developed such as Eigen Planarity, Voxel Density, Voxel DeltaZ, WENS Z. The combination of these parameters is composed of an algorithm to detect misclassified DTD area. The experiments show the proposed algorithm is independent of the LiDAR data density and to minimize the effects of topographic relief. In other words, the result was so reliable. I think the proposed method contribute to LiDAR filtering process improvement.

Abstract LiDAR

LiDAR 3,,. (DEM), GIS DEM (X. Meng, 2010). 1:5,000 (, 2009), (2007) LiDAR /. DEM LiDAR 1), 2). 1.

LiDAR (G. Vosselman, 2004),.,. LiDAR, (Surface) LiDAR., LiDAR,. G. Vosselman (2004) 8, 15.,,.,,. P. Axelsson (2000) Seed point TIN, TIN Adaptive-TIN Densification.,,. LiDAR TerraScan (TerraSolid, 1999). (2006).

, (Blunder),. LiDAR Slope-based (G. Vosselman, 2001) K. Zhang (2003) Progressive morphology. K. Zhang(2003) mathematical morphology filter,.,. LiDAR,. G. Sithole (2000) Uncertainty( )., XY (String). (Score), 0, 1 0.5 Uncertainty. Uncertainty, (Height difference). LiDAR. (2007) 3., 1 Morphological Filter. 2,. 3 2, 1.. (2007) LiDAR.,

,. DEM. (2007) LiDAR,.., LiDAR. GIS (Automation) (Accuracy)... LiDAR., LiDAR. LiDAR, LiDAR.

LiDAR (DTD). (Voxel-based Approach), LiDAR. 2. 2.

, 6. 1 : LiDAR. 2 : LiDAR 3 3D,. 3. 3 : Eigen Planarity, 1. 4 : Voxel Density,.. 5 : Voxel DeltaZ,.. 6 : WENS Z,.

LiDAR DTD,. ( 2009-950 ). (Mass points) 3. LiDAR,,,,,,.,. 4. 4. LiDAR

,, 1/5000, 1/1000. Point Data Record Format 1 (Las 1.0), X(East), Y(North), Z(Height), Intensity, Return_Number, Number_of_Returns, GPS_Time, Classification_Code. (Ground) 2, (Unclassified) 1,,,,,. ) C001_E 090101.las ( : E) C001_O 090101.las ( : O) 1.

LiDAR..,.,,.. LiDAR 3. (Point-cloud)., 2 3. (Point-to-point) 2.,.. (Point-to-points).,.

(Points-to-points).. 2. (ISPRS, 2003). 3,,, /. (Slope).,.. (Block Minimum),.,.

(Surface Modeling),. / (Clustering/Segmentation). ( ). 3. (ISPRS, 2003)

LiDAR., LiDAR.. (X. Meng, 2010) (Bridge) 5. (X. Meng, 2010)

,.,. Morphology,. LiDAR S/W Terra solid (www.terrasolid.fi) (Terra scan). Classify G round 10% (F. Crosilla, 2004),.,. LiDAR, (G. Vosselman, 2004).,.,.

DEM,,. LiDAR (Commission Error) (Omission Error),,. (Type- Error), (Type- Error), (Total Error). Cross-matrix., (G. Sithole, 2004).,., DEM..

,,., (X. Meng, 2010).. LiDAR TerraScan (Max Building Size). 2,. TIN,. 6.,..,.

,.,,.. DEM, 8.

.,.... LiDAR,.,. 9 10.

,. (Outlier), 11 DEM..

, 12. DEM.. (Outlier) DEM.

.,. Volumn (Voxel),. LiDAR. GIS(Geographic Information System),, (Raster). 3 TIN(Triangulated Irregular Network)., (Rendering),. (J. Stoker, 2009) (Voxel), 3. (Modeling), (Unit). LiDAR., (Waveform) LiDAR, (D. Van, 2009), (J. Reiberger, 2008).

3 (Regular 3D Grid),. (Sub-voxel), 13 (Cube). XYZ, (index). 3,. (1). (, 2005) (1)., 3. LiDAR., 5m, (Powerline) (2009) 1m 1m.

.,,, 14.

14 25m ~ 30m.,.. 15 10m ~ 70m.

15, 10m 15 (a). (b) 30m, (c) (d) 50m.,., 30m. 16 LiDAR 30 x 30 x 30(m).

LiDAR LiDAR.,. Eigen Planarity LiDAR (eigenvalue).. (linear algebra), LiDAR (xyz). LiDAR (Basis), S = {,,, } V. LiDAR 17 3. 17. 3 LiDAR (Random Vector) (Covariance Matrix),

LiDAR. LiDAR 3 ODR(Orthogonal distance Regression). LiDAR (Centeroid), 18. 18. LiDAR. (2) M, (2) (3). (3)

LiDAR (3), 3 x 3., 5 LiDAR. 5 LiDAR. (4) Eigen Planarity. (4)

Eigen Planarity (Axis) LiDAR. (4) 1st, 2nd, 3rd,. 19 Eigen Planarity. 19. Eigen Planarity 19 LiDAR, A, B, C Eigen Planarity (= 1).,. D, E, F. (translation), (rotation) (scaling). 19 LiDAR. 20 LiDAR Eigen Planarity.

20 LiDAR 0.9. LiDAR Z. Z (4) 3rd Eigenvalue, Eigen Planarity.. 20, 1st 2nd Eigen Planarity. 3rd Eigenvalue', Eigen Planarity 3.. Z, Eigen Planarity.

21 Eigen Planarity, A, B (Bridge). Eigen Planarity 0.9. 1 Eigen Planarity 0.9. Eigen Planarity 22 A B, C. Eigen Planarity,. Eigen Planarity,. 6 Eigen Planarity. double Voxel_compute_Planarity (input Eigenvalues) /* Voxel 고유값 : Eigenvalue 1 > Eigenvalue 2 > Eigenvalue 3 */ double Eigenvalue 1 double Eigenvalue 2 double Eigenvalue 3 for (1st Voxel; Last Voxel; Voxel++) { Planarity = (Eigenvalue 2 - Eigenvalue 3) Eigenvalue 1 } return Planarity

Voxel Density. ( ) Eigen Planarity,., 23 LiDAR. LiDAR (Density) xy (1 ). Voxel Density 3 2 (1 ). (5) Voxel Density. 24, 24.

Voxel Density 2/3,. (6) Voxel Density. (6) 24, (6)..,.,, (6) 2/3. 25 Eigen Planarity, Voxel Density.

25 Eigen Planarity. Voxel Density (D Density) (V Density),. 26,. Voxel Density. Voxel Density,. 7 Voxel Density. double Voxel_compute_Density for (1st Point; Last Point; Point++) { /* Voxel 내 LIDAR 포인트갯수 */ The Number of Points = Add (Point) } Voxel Density = Number of Points 900m ( = 30 30) return Voxel Density

Voxel DeltaZ Eigen Planarity Voxel Density, LiDAR Z. Eigen Planarity, Voxel DeltaZ. Z, Voxel DeltaZ (7). (7) 27 Voxel DeltaZ. Voxel DeltaZ 10%, 20%. 30m 0.1( = ), 0.2( = ). (8) Voxel DeltaZ. (8) Eigen Planarity Z,. 28 LiDAR.

29 Eigen Planarity Voxel Density. Voxel DeltaZ 0.2,. 30 Z.

31 Eigen Planarity Voxel Density. 31 Voxel DeltaZ. 31 0.2, 0.1. Voxel DeltaZ Eigen Planarity,. LiDAR,. 32.

32, Z Voxel DeltaZ. 8 Voxel DeltaZ. double Voxel_compute_Delta Z for (1st Point; Last Point; Point++) { /* 모든 LiDAR 포인트 Z 값탐색 Voxel 내최대 Z 값포인트선택 */ Highest Z } /* Voxel 내최소 Z 값포인트선택 */ Lowest Z Voxel Delta Z = 절대값 Highest Z - Lowest Z Voxel size return Voxel Delta Z WENS Z Eigen Planarity, Voxel Density, Voxel DeltaZ (, )., WENS Z. 33,.

LiDAR LiDAR, (9). (9) West Z, East Z, North Z, South Z LiDAR. 34,. 34.,. WENS Z Eigen Planarity, Voxel Density, Voxel DeltaZ, (10) WENS Z.

(10) LiDAR WENS Z,. 9 WENS Z. double Voxel_compute_WENS Z for (1st Voxel; Last Voxel; Voxel++) { /* 대상 Voxel내포인트의평균 Z 값 */ Voxel Z } /* 동서남북인접 Voxel의평균 Z 값 */ W(West) Voxel Z E(East) Voxel Z N(North) Voxel Z S(South) Voxel Z WENS Z = (W + E + N + S) / 4 if (Voxel Z > WENS Z) Detect Commission Error else Detect Omission Error

35., Eigen Planarity Voxel Density, Voxel DeltaZ WENS Z.

. 1 : LiDAR. 2 : LiDAR 3 3D,. 3 : Eigen Planarity, 1. 4 : Voxel Density,.. 5 : Voxel DeltaZ,.. 6 : WENS Z,.. LiDAR 36.

36,., WENS Z. 37. Eigen Planarity, Voxel DeltaZ, WENS Z,. 38. LiDAR, 38 (a) 1, 3, 4 (b) 4, 5, 6, 7. 3,

Eigen Planarity. 38 (a) 2 3,. 6 Voxel DeltaZ,. Eigen Planarity 1., 39. 39,, Eigen Planarity, Voxel DeltaZ, Voxel Density, WENS Z.

., DEM. LiDAR,. LiDAR. 40,. 40.,. LiDAR, DEM.

MS Visual Studio.NET 2005 3D OpenGL(Open Graphics Library)., 3. 1 :. 2 :. 3 :. LiDAR. 1,.. LiDAR. LiDAR,. 41. 5.4, 2.7, 1.3 double type, 6.

41 Eigen Planarity, Z DeltaZ, WENS Z., LiDAR. Voxel Density,.. LiDAR ( ), 10.

42 LiDAR. 42, LiDAR., 43.

43 LiDAR 0.5,. 1, 2 3. 44.

44 30 30 30(m),..,. 45.

LiDAR, 46 LiDAR. 46,,. 47.

48,.

. LiDAR ( ), 11. 49 LiDAR,.,.

,. 0.5,. 50,.,.. 51.

..,.

( ), 12. 52 LiDAR.

,. 52. 53.,. 53,,.. 53,.,.

LiDAR. 3D LiDAR.., LiDAR.,,., LiDAR., LiDAR.,.., /,.., LiDAR. LiDAR,..

, LiDAR.,. LiDAR,., LiDAR. LiDAR. LiDAR.