The 3-D reconstruction from sequence of images finds many applications in modern computer vision systems such as virtual reality, vision-guided surgeries, autonomous navigation, medical studies and simulations, reverse engineering, and architectural design. The very basic requirement of these applications is to find accurate and realistic reconstructions. In fact, the 3-D scene reconstruction from multiple images is a challenging and interesting problem to tackle. It is interesting because humans naturally solve this problem in an easy and efficient way. However, it is a challenge because there is no single solution of many different solutions proposed to solve the problem has the completeness of the human’s solution. Of course, there are good solutions and there may be others in the future. This project has an ultimate goal of guiding the research of 3-D model building towards better performance of image-based 3-D reconstruction techniques. To achieve this goal we introduce a unified framework for performance evaluation of 3-D reconstruction techniques from sequence of images. This framework provides designs and developments of the following building blocks: (1) Evaluation testbed, (2) Pre-evaluation methodologies, (3) Performance evaluation strategies, and (4) Applications (Post-evaluation).
Research Team Aly A. Farag CVIP Director farag@cvip.louisville.edu Ahmed H. Eid Research Associate eid@cvip.louisville.edu 
Methods
1- Performance Evaluation Testbed
To provide the input sequence of images to the vision technique under-test and the ground truth data necessary to examine the performance of the given vision technique. With this setup we are able to build database of ground truth data and intensity data as well to be ready for use by the vision community for further studies of performance tracking and analysis of different 3-D reconstruction techniques.
2- Pre-evaluation Methodologies
to develop techniques for preparing the data under-test and the ground truth data for the following evaluation steps. For example, background subtraction and 3-D data registration through silhouettes (RTS) techniques.
3- Performance Evaluation Strategies
to develop performance evaluation strategies and measuring criteria to quantify the performance of the given techniques under-test. For example, local quality assessment (LQA) technique
4- Post-evaluations (Applications)
to develop methods for analyzing the evaluation results for diagnosis purposes. Further steps include data fusion in competitive-cooperative fashion. For example, the closest contour (CC) technique.
Results
1- Performance Evaluation Testbed
b) Top View
a) System Setup
2- Pre-evaluation Methodologies
(a) Background Subtraction
(b) 3-D Data Registration
3- Performance Evaluation Strategies
where Pq is the probability estimate of quality
4- Post-evaluations (Applications)
Movies
Stereo Cyclograph(14.5 MB) and Zipped Version (7.5 MB) Scanning (9.8 MB) and Zipped Version (5.2 MB) Registration Results_view0 (1.28 MB) and Zipped Version (693 KB) Registration Results_view90 (942 kB) and Zipped Version (655 KB) Fusion (2.75 MB) and Zipped Version (1.75 MB)
Publications
1- A.A. Farag and A.H. Eid, “On the performance evaluation of 3-D reconstruction techniques from a sequence of images,” EURASIP Journal on Applied Signal Processing, 2005. 2- A.A. Farag and A.H. Eid, “Silhouette-Contour based 3-D registration methodology as a pre-evaluation step of 3-D reconstruction techniques,” IEEE International Conference on Image Processing (ICIP05), 2005. 3- A.H. Eid and A.A. Farag, Design of an Experimental Setup for Performance Evaluation of 3-D Reconstruction Techniques from Sequence of Images, Proceedings of Applications of Computer Vision Workshop in conjunction with the European Conference on Computer Vision (ECCV04), pp. 69-77, Prague, May 2004. 4- A. Eid and A.A. Farag, A unified framework for performance evaluation of 3-D reconstruction techniques, Proc. of IEEE Conference on Computer Vision and Pattern Recognition (CVPR’04), Workshop on Real-time 3-D Sensors and their Use, Washington, DC, June 27-July 2, 2004, pp. 41-48. 5- A.H. Eid and A.A. Farag, On the fusion of 3-D reconstruction techniques, Proc. of Seventh International Conference on Information Fusion (Fusion-04), Stockholm, Sweden, June 28-July 1, 2004, pp. 856-861. 6- A.A. Farag and A.H. Eid, Local quality assessment of 3-D reconstructions from sequence of images: a quantitative approach, Proc. of Advanced Concepts for Intelligent Vision Systems (ACIVS’04), Brussels, Belgium, August 31-September 3, 2004, pp. 161-168.
Acknowledgement / Sponsors
This work is supported by US Army under grant DABT60-02-P-0063 and Air Force Office of Scientific Research (AFOSR) grant F49620-01-1-0367.
Copyright © 2011, All rights reserved to CVIP Laboratory, Computer Vision & Image Processing Laboratory, University of Louisville,Kentucky, USA.
Designed & Maintained by webmasters team
CVIP Lab 