3 : RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) (Special Paper) 24 5, 2019

Size: px

Start display at page:

Download "3 : RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) (Special Paper) 24 5, 2019"

지혜 팽
4 years ago
Views:

1 3: RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) (Special Paper) 24 5, (JBE Vol. 24, No. 5, September 2019) ISSN (Online) ISSN (Print) RGB-D a), a), a), a) Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects Kyung-Jin Kim a), Byung-Seo Park a), Dong-Wook Kim a), and Young-Ho Seo a) RGB-D.. 3D 3D Camera. 2. RGB-D 8 3. Abstract In this paper, we propose a point cloud matching algorithm for multiple RGB-D cameras. In general, computer vision is concerned with the problem of precisely estimating camera position. Existing 3D model generation methods require a large number of cameras or expensive 3D cameras. In addition, the conventional method of obtaining the camera external parameters through the two-dimensional image has a large estimation error. In this paper, we propose a method to obtain coordinate transformation parameters with an error within a valid range by using depth image and function optimization method to generate omni-directional three-dimensional model using 8 low-cost RGB-D cameras. Keyword : Registration, RGB-D Camera, Point Cloud, 3D, Camera Pose Estimation a) (Kwangwoon University Electronic Materials Engineering) Corresponding Author : (Young-Ho Seo) yhseo@kw.ac.kr Tel: ORCID: (R _ ). (This research was supported by Cultural Technology R&D Project through the Korea Creative Content Agency(R _ )). Manuscript received July 29, 2019; Revised September 19, 2019; Accepted September 19, Copyright 2016 Korean Institute of Broadcast and Media Engineers. All rights reserved. This is an Open-Access article distributed under the terms of the Creative Commons BY-NC-ND ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited and not altered.

2 (JBE Vol. 24, No. 5, September 2019).,,,,, (Virtual Reality:VR), (Aug- mented Reality: AR), (Mixed Reality: MR) Microsoft research 2011 Kinect- Fusion [1] RGB-D 3 [2]. RGB-D 3 [2].. ICP (Iterative Closest Point) [3].,. ICP SoftAssign [4] [5][6][7] [8]., (local minima) [8]. PCA(Principal Component Analysis) alignment, modal spectral matching [8][9][10]. [8] 8 RGB-D. Zhang [11]. [12], RGB 3 [2]. ICP RGB RGB D RGB RGB-D. 3D 8 RGB-D. 3 RGB-D RGB. 8 RGB-D. 4,,. 1. 1(a) 1(b)

3 3: RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) (a) 1. 3D (a), (b) Fig. 1. 3D Point Cloud Shooting System (a) vertical, (b) horizontal photographing angle and range (b). RGB-D 3. [13].,., [14] RGB.. [15]. 2 RGB-D 3.. RGB RGB, Fig. 2. Actual-based 3D Mesh Model Sequence Generation Flowchart

(JBE Vol. 24, No. 5, September 2019) RGB-D 3 RGB... 3. 1.

4 (JBE Vol. 24, No. 5, September 2019) RGB-D 3 RGB QR (Charuco board) [16]. QR board [11][16]. (2D) (3D) [11][17]... Charuco board, x, y x y z. [18]. 3. 3(a) 3(b). [16]. (1). R t. R t (1) [19].... (1), (2) (3). (a) (b) 3. (a), (b) Fig. 3. (a) Charuco board, (b) World coordinate system obtained through Charuco board

5 3: RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) ( ). (6). (3) (4). (4) (2). (5). ( ) ( ) (gradient descent), (gradient descent with momentum), (Nesterov s gradient method) [20].. x, y, z, 6.. (2)~(5). (Squared Euclidean Distance, SED). (7). (8) RGB-D RGB. RGB "Charuco Board Corner Detection" (1). (2)~(5) "Initial Coordinate Transform Parameter Calculation". "Noise Reduction", 3 "Charuco Board Corner 3D Coordinate Calculation". 3 (6). "Iterative Operation for Minimum SED" (7) (8).. 4

6 (JBE Vol. 24, No. 5, September 2019).. 8 Intel RealSense D m 4 1.7m. 4. Fig. 4. Parameter optimization Flowchart 5. Fig. 5. Camera system (a) (b) (c) (d) (e) (f) (g) (h) 6., (a), (b), (c), (d), (e), (f), (g), (h) Fig. 6. Point cloud acquired from each camera, (a) front, (b) left, (c) right, (d) rear of the cameras placed above, (e) front, (f) left, (g) right, (h) rear of the placed cameras below

7 3: RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) 0.1m 1.5m RGB-D (a), (b), (c), (d),,, 6(e), (f), (g), (h). Charuco board , , , Camera1 Camera1. 59cm, 3mm Table 1. Average registration error of point cloud Initial Parameter(cm) Parameters applied with optimization(mm) Camera Average x. RGB-D RGB RGB 3 7. Fig. 7. Graph of average error distance according to iteration

(JBE Vol. 24, No. 5, September 2019) (a) (b) (c) (d) (e) (f) 8.

The omnidirectional point cloud before and after the optimization algorithm (a) front, (b) left, (c) back of

(d) front, (e) left, (f) back of the point cloud with optimization parameters x. 3. 8 RGB-D. 8. 8(a), (b), (c),,.

8 (JBE Vol. 24, No. 5, September 2019) (a) (b) (c) (d) (e) (f) 8., (a), (b), (c), (d), (e), (f) Fig. 8. The omnidirectional point cloud before and after the optimization algorithm (a) front, (b) left, (c) back of the point cloud with pre-optimization parameters. (d) front, (e) left, (f) back of the point cloud with optimization parameters x RGB-D. 8. 8(a), (b), (c),,. 8(d), (e), (f),,.. Zhang 587mm, 2.9mm. 200., RGB Charuco board.. RGB-D 3. RGB (References) [1] S. Izadi, D. Kim, O. Hilliges, D. Molyneaux, R. Newcombe, P. Kohli, J. Shotton, S. Hodeges, D. Freeman, A. Davison, A. Fitzgibbon, KinectFusion: Real-Time Dynamic 3D Surface Reconstruction and

3: RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) Interaction Using a Moving Depth Camera ISMAR, Vol. 11, No. 2011, pp.

309-314, 2014. [3] C. Gregory, W. Sang, K. David, ICP Registration Using Invariant Features., IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, Vol. 24, No. 1, pp. 90-102, January 2002.

9 3: RGB-D (Kyung-Jin Kim et al.: Point Cloud Registration Algorithm Based on RGB-D Camera for Shooting Volumetric Objects) Interaction Using a Moving Depth Camera ISMAR, Vol. 11, No. 2011, pp , October [2] S. Choi, S. Park, Convenient View Calibration of Multiple RGB-D Cameras Using a Spherical Object. KIPS Transactions on Software and Data Engineering, Vol.3 No.8, pp , [3] C. Gregory, W. Sang, K. David, ICP Registration Using Invariant Features., IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, Vol. 24, No. 1, pp , January [4] S. Gold, A. Rangarajan, C. Lu, S. Pappu, E. Mjolsness, New algorithms for 2d and 3d point matching: Pose estimation and correspondence. In Pattern Recognition, Journal of KIISE, Vol. 31, pp , August [5] S. Granger, X. Pennec. Multi-scale em-icp: A fast and robust approach for surface registration. In ECCV, pp , June [6] P. David, D. DeMenthon, R. Duraiswami, H. Samet. Simultaneous pose and correspondence determination using line features. In CVPR, pp , June [7] W. Zhao, D. Nister, S. Hus. Alignment of continuous video onto 3d point clouds. PAMI, Vol. 27, pp , August [8] Li, Hongdong, R. Hartley. "The 3D-3D Registration Problem Revisited." 2007 IEEE 11th International Conference on Computer Vision, pp. 1-8, 2007 [9] S. Sclaroff, A. Pentland. Modal matching for correspondence and recognition. In PAMI, June [10] M. Leordeanu, M. Hebert. A spectral technique for correspondence problems using pairwise constraints. In ICCV, Vol. 2, pp , 2005 [11] Z. Zhang, A flexible new technique for camera calibration, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.22, No.11, pp , [12] J. Lee, Camera calibration and Compensation distortion, Korea robotics society review, Vol. 10 No, 1, pp , February 2013 [13] B. Lin, M. Su, P. Cheng, P. Tseng, S. Chen Temporal and Spatial Denoising of Depth Maps, Sensors 2015 Vol. 15, No. 8, pp , August [14] J. Digne, C. Franchis, The Bilateral Filter for Point Clouds, Image Processing on Line, Vol. 2017, No. 7, pp , March [15] T. Fang, L. A. Piegl. "Delaunay triangulation in three dimensions." IEEE Computer Graphics and Applications, Vol. 15, No. 5, pp , September 1995 [16] Open Source Computer Vision, tutorial_charuco_detection.html [17] Open Source Computer Vision, tutorial_aruco_calibration.html [18] Q. Zhang, R. Pless, Extrinsic calibration of a camera and laser range finder (improves camera calibration)., In 2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)(IEEE Cat. No. 04CH37566), Vol. 3, pp , September [19] P. Rathnayaka, S. Baek, S. Park, Calibration of a Different Field-of-view Stereo Camera System using an Embedded Checkerboard Pattern, International Conference on Computer Vision Theory and Applications, pp , 2017 [20] S. Lee, Convergence Rate of Optimization Algorithms for a Non-strictly Convex Function, Institute of Control Robotics and Systems, pp , May : () ~ : () - ORCID : - :,, 2D/3D : ~ : () - ORCID : - :,, 2D/3D

10 (JBE Vol. 24, No. 5, September 2019) : () : : Georgia () ~ : - ORCID : - : 3D,, VLSI Testability, VLSI CAD, DSP, Wireless Communication : () : () : () ~ : ~ : - ORCID : - :, 2D/3D,

09권오설_ok.hwp

09권오설_ok.hwp (JBE Vol. 19, No. 5, September 2014) (Regular Paper) 19 5, 2014 9 (JBE Vol. 19, No. 5, September 2014) http://dx.doi.org/10.5909/jbe.2014.19.5.656 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) a) Reduction