ShapeLab: A Unified Framework for 2D & 3D Shape Retrieval

  • Authors:
  • Jiantao Pu;Karthik Ramani

  • Affiliations:
  • Purdue University, USA;Purdue University, USA

  • Venue:
  • 3DPVT '06 Proceedings of the Third International Symposium on 3D Data Processing, Visualization, and Transmission (3DPVT'06)
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

2D or 3D shapes are the most important visual information that we use to recognize an object. We propose a unified framework "ShapeLab" to search similar 2D or 3D shapes from an existing database. Users can search 3D shapes with a 2D input, and vice versa. ShapeLab is composed of four key components: (1) pose determination for 3D models; (2) 2D orthogonal view generation based on multiple levels of detail; (3) similarity measurement between 2D shapes; and (4) freehand sketch-based user interface. Key algorithms supporting the above components are briefly described. Experiments show ShapeLab can provide a better performance such as high accuracy, flexibility and scalability compared to the available methods.