利用單一Kinect實現即時之物件三維重建

Real-time Reconstruction of 3D Objects Using a Single Kinect

鄭惟浩(Wei-Hao Cheng)

指導教授 ： 顏嗣鈞

國立臺灣大學/電機資訊學院/電機工程學研究所/碩士(2014年)

https://doi.org/10.6342/NTU.2014.02886

摘要

在Microsoft Kinect問世之後，深度的資訊處理已被廣泛的研究，進而產生三維重建此一熱門議題，常被應用於醫學、考古學、藝術品建模等等。本論文透過Kinect攝影機之深度與色彩資訊擷取，經由使用者手持Kinect對物件進行環繞拍攝，系統可即時建構出所拍攝物件之彩色紋理三維模型。利用物件與所在平面之深度值差與深度斜率變化來偵測物件邊緣，並將畫面中之線條配合EmguCV尋找出多個輪廓，經由深度資訊與輪廓大小之門檻，來判斷物件輪廓的所在位置，進而將物件輪廓內之深度值與色彩資訊配合KinectFusion，即可得到物件的三維紋理模型，使用者可即時的觀察場景與物件之三維重建過程。以本論文之系統與Kinect RGBDemo v0.7.0之程式做比較，本篇論文之環境設置較為簡單，且物件重建的結果也較為精緻。

關鍵字

三維重建 ； 邊緣偵測 ； 輪廓 ； KinectFusion ； 紋理模型

並列摘要

Since the debut of the Microsoft Kinect in 2010, depth information has been extensively applied to a variety of areas in academic research and industrial products. Among areas in which depth information plays a key role, 3D reconstruction has become a popular issue, with applications in medical imaging, archeology, and art reconstruction. With a user holding a Kinect camera which captures RGB and depth streams and moving around the target object, a textured 3D model can be reconstructed in real-time. The main purpose of this research is to 3D reconstruction in real-time using a single Kinect. To this end, we first define appropriate depth thresholds to perform the edge detection. Then we use EmguCV method findcontours to check for contours. Finally, we reconstruct a textured model from data of contours with KinectFusion. Experimental results are compared against the results of an existing system, showing the accuracy and effectiveness of the proposed design.

並列關鍵字

3D Reconstruction ； Edge Detection ； Contours ； KinectFusion ； Textured Model

參考文獻

[1] W. Matusik, et al., "Image-based 3D photography using opacity hulls," ACM Transactions on Graphics, Vol. 21, No. 3, pp. 427-437, July 2002.

[2] M. Ribo, M. Brandner, "State of the art on vision-based structured light systems for 3D measurements," IEEE International Workshop on Robotic and Sensors Environments, pp. 2-6, 2005.

[3] H. Schafer, F. Lenzen, and C. Garbe, "Depth and intensity based edge detection in time-of-flight images," IEEE International Conference on 3DTV-Conference, pp. 111-118, June 2013.

[4] W. Chen, H. Yue, J. Wang, and X. Wu, "An improved edge detection algorithm for depth map inpainting," Optics and Lasers in Engineering, Vol. 55, pp. 69-77, April 2014.

[7] S. Izadi, et al., "KinectFusion: real-time 3D reconstruction and interaction using a moving depth camera," in Proceedings of the 24th annual ACM symposium on User Interface Software and Technology, pp. 559-568, October 2011.