Past Research Projects
Here you find a partial list of my previous projects (this page has not been updated for a long time):
Relief Texture Mapping
Please check the Relief Texture Mapping web site for a comprehensive description of the project and list of publications.
We are building a portable, low-cost, and easy-upgradable imaging system for videolaparoscopy. Our system will be completely digital and will provide better image resolution than commercial products. Another unique feature of this system is the ability to process the video stream in real-time in order to enhance details.
Our system will cost just a fraction of the price of similar commercial products and should make videolaparoscopy, and videosurgery in general, more affordable.
This project is sponsored by CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico).
Real-Time Video and Image Metrology
We are developing new computer vision and image processing algorithms and building scanners for making measuments directly from images. Our goal is to be able to make reliable measurements of some scene properties, such as the volumes of simple objects, in real time.
Video Illustrating the Technique
Related publications:Leandro A. F. Fernandes. Um método projetivo para cálculo de dimensões de caixas em tempo real. Master's Thesis under supervision of Prof. Manuel M. Oliveira. PPGC-UFRGS, January, 2006. (in Portuguese). (First place in the 2007 Brazilian Computing Society's Masters Thesis contest)
Leandro Fernandes and Manuel M. Oliveira. Real-time Line Detection Through an Improved Hough Transform Voting Scheme. Pattern Recognition, Elsevier. Volume 41, Number 1, January 2008. pp. 299-314. (ISSN 0031-3203).
Leandro Fernandes, Manuel M. Oliveira, Roberto da Silva and Gustavo Crespo. A Fast and Accurate Approach for Computing the Dimensions of Boxs from Single Perspective Images. Journal of the Brazilian Computer Society (JBSC), Number 2, Vol. 12, September 2006, pp. 19-30. (ISSN 0104-6500).
Leandro Fernandes and Manuel M. Oliveira A Scanner for Computing Box Dimensions in Real Time. ACM SIGGRAPH 2006 Conference Abstracts and Applications. Available in the Conference CDROM (ISBN 1-59593-366-2).
Leandro A. F. Fernandes., Manuel M. Oliveira and Roberto da Silva and Gustavo Crespo. Computing Box Dimensions from Single Perspective Images in Real Time. Proceedings of the XVIII Brazilian Symposium on Computer Graphics and Image Processing (SIBGRAPI 2005). pp. 155-162. (ISBN 0-7695-2389-7).
Crespo, Gustavo J. Computing Box Volumes from Images: An Implementation. Masters Project, SUNY at Stony Brook, July 2002.
Oliveira, Manuel M. Computing Box Volumes from Single Images. SUNY Stony Brook Technical Report TR01.10.31, State University of New York at Stony Brook, October 31, 2001.
The goal of this project is to build an inexpensive and portable 3D camera comparable in size and easy of operation to a consumer-grade digital camera.
Our first prototype, based on a monochrome VGA-resolution camera, is capable of recovering the 3D shape of smooth objects, such as human faces, from a single picture. The first prototype was developed as part of Askold V. Strat's Masters Thesis ("A Portable and Inexpensive Laser-Based 3D camera", July 2002) at SUNY Stony Brook. For a concise description of this project An Inexpensive 3D Camera (ACM SIGGRAPH 2002 Conference Abstracts and Applications, page 146).
Our second prototype (shown on the left) is comprised of three main hardware components: a digital camera (Olympus C-2020 Zoom), a laser pattern (raster) generator and an interface between the digital camera and the raster generator. Some image processing software is used to reconstruct shapes from the acquired images. The raster generator and the interface are attached to the camera using the tripod screw hole and a bolt/nut combination, which provides a rigid and consistent mounting. Conceptually, the system works as follows: the raster generator projects a set of laser planes in the scene, each at a different elevation angle. The camera acquires a pair of pictures with a single press of the camera’s trigger. One of the pictures contains a set of projected laser lines (created by the rastergenerator) and is used to reconstruct the object’s imaged geometry using triangulation. The other picture is used as texture. Below, you see a pair of pictures acquired by the 3D camera (left and center) and the reconstructed 3D model for the face (shown on the right).
Publications related to our 3D Color Camera:
Askold Strat and Manuel M. Oliveira A Point-and-Shoot Color 3D Camera. Proceedings of the 4th International Conference on 3-D Digital Imaging and Modeling (3DIM), Banff, Canada, October 6-10, 2003, pp. 483-491.
Askold Strat and Manuel M. Oliveira Casual 3D Photography. ACM SIGGRAPH 2003 Conference Abstracts and Applications.
Modeling and Rendering of Real Environments
We are developing new algorithmic solutions for model reconstruction of real environments. These solutions compensate for the limitations of 3D scanning techniques, such as occlusions and accessibility limitations. More details can be found in the following publications:
Jianning Wang and Manuel M. Oliveira Improved Scene Reconstruction from Range Images. Computer Graphics Forum Volume 21 (2002) Number 3, pp. 521-530. Proceedings of Eurographics 2002.
Jianning Wang and Manuel M. Oliveira Filling Holes on Locally Smooth Surfaces Reconstructed from Point Clouds. Image and Vision Computing, Elsevier. Volume 25, Issue 1, 1 January 2007. pp. 103-113. (ISSN 0262-8856).
Corrêa, Wagner, Manuel M. Oliveira, Cláudio T. Silva and Jianning Wang. Modeling and Rendering of Real Environments. RITA - Revista de Informática Teórica e Aplicada, Volume IX, Number 2, pp. 127-156, October 2002.
Jianning Wang and Manuel M. Oliveira. A Hole Filling Strategy for Reconstruction of Smooth Surfaces in Range Images. XVI Brazilian Symposium on Computer Graphics and Image Processing. São Carlos, SP. October 12-15, 2003, pp. 11-18.
Surface Reconstruction from Point Clouds
We are developing new algorithms for surface reconstructing from point clouds. We are particularly interested in algorithms for reconstructing non-manifold surfaces and surfaces with boundaries, as well as for performing reconstruction from noisy datasets (when normals cannot be reliably estimated).
We have built some software pipeline for surface reconstruction of complex geometric objects from point clouds. This effort is complemented by the construction of a 3D laser scanner prototype (under development). The next steps will include the capture of optical properties of object surfaces and new paradigms for editing implicit representations.
Jianning Wang, Manuel M. Oliveira and Arie Kaufman. Reconstructing Manifold and Non-Manifold Surfaces from Point Clouds. Proceedings of IEEE Visualization 2005, Minneapolis, MN, October 23-28, 2005. pp. 415-422
Jianning Wang, Manuel M. Oliveira, Hui Xie and Arie Kaufman. Surface Reconstruction Using Oriented Charges. Computer Graphics International 2005, Stony Brook, NY, June 22-24, 2005, pp. 122-128.
Manuel M. Oliveira and Guilherme Parisotto Guimarães. Reconstructing Implicit Surfaces with Boundaries. UFRGS Technical Report, July 2004.
to my home page