3D Object Modeling with RGB-D images


   Jun Xie1    Mayoore Jaiswal1    Rogerio Feris2    Ming-Ting Sun1  
1University of Washington      2IBM Research



RGB-D (Kinect-style) cameras are novel low-cost sensing systems that capture RGB images along with per-pixel depth information. In this paper we investigate the use of such cameras for acquiring multiple images of an object from multiple viewpoints and building complete 3D models of objects. Such models have applications in a wide range of industries. We implemented a complete 3D object model construction process with object segmentation, registration, global alignment, model denoising, and texturing, and studied the effects of these functions on the constructed 3D object models. We also developed a process for objective performance evaluation of the constructed 3D object models. We collected laser scan data as the ground truth using a Roland Picza LPX-600 Laser Scanner to compare to the 3D models created by our process.
 
Given the RGB-D images of an object from different views, the basic 3D object modeling process includes:
1. SIFT/SURF feature based initial alignment
2. ICP fine registration
3. ELCH based global alignment
3. Model fitting and noise reduction

Here are some short demos for the 3D object modeling:

Demo 1
  
Demo 2


[Source Code] (Github)


This package is designed to use RGB-D images for 3D object reconstruction. If you use this code for your reasearch, please cite the following papers:

[1] J. Xie, Y.F. Hsu, R. Feris, M.T. Sun, "Fine registration of 3D point clouds fusing structural and photometric information using an RGB-D camera," 
Journal of Visual Communication and Image Representation (JVCI), vol. 32, pp. 194-204, Oct. 2015. [PDF]

[2] M. Jaiswal, J. Xie and M.T. Sun, "3D Object Modeling with a Kinect Camera," 
Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) 2014 [PDF]