Ricardo Fabbri

Basic Info

PhD Candidate in Computer Engineering, at LEMS, Brown University, sponsored by CNPq.
Advisor: prof. Benjamin Kimia
Google, Inc. Intern, Book Search Team, May-Aug 2008.
Master in Computer Science, USP at Sao Carlos, Brazil
Bsc. Computer Science, USP at Sao Carlos, Brazil

Research abstract

I currently work with computer vision using multiple images of the same scene (as acquired by a camera in different positions, systems of multiple cameras, or video). This is useful for automating recognition, location, and measurement of 3D objects from images, 3D photography, etc. I am concerned with devising robust, precise and automatic methods for tackling problems in this area. In order to meet such requirements, I explore the use of grouped primitives such as curves and their differential geometry as the basis for the methods. I believe this is a much more powerful approach to solve multiview problems than the traditional one which relies on point primitives. I have been recently working on fusing a number of cues for 3D reconstruction. The end-goal is a system based on a hand-held video sequence, without need for calibration or textured regions, being able to identify not only the 3D structure of objects, but also the camera position, the reflectance properties, and lighting conditions as well. More information is provided in the following link:

Multiview Differential Geometry of Curves and Surfaces

Publications

R. Fabbri, O. M. Bruno, J. C. Torelli, L. da F. Costa "2D Euclidean Distance Transform Algorithms: A Comparative Survey", ACM Computing Surveys, Feb 2008. (pdf | bib | errata)
- Supplementary material and source code: distance.sf.net
R. Fabbri, B. B. Kimia, "High-Order Differential geometry of Curves for Multiview Reconstruction and Matching", EMMCVPR 2005, Lecture Notes in Computer Science, 3757, pp. 645-660, 2005. (original | corrected version)
- Theoretical paper on recovering torsion and curvature derivative of a space curve from multiple perspectives
- The formulas in the original paper have some minor imprecisions regarding the sign of the curvature. The corrected version fixes this issue.
- I wrote extensions of the formulas that take into account the intrinsic parameters. They show how the image point, tangent, curvature, and curvature derivative change under the transformation given by the intrinsic parameter matrix K.
- I implemented the formulas of this paper in a curve reconstruction system, just drop me an email and I will provide you with the code.
R. Fabbri, L. F. Estrozi, L. da F. Costa, "On Voronoi Diagrams and Medial Axes", Journal of Mathematical Imaging and Vision, 17(1), 27-40, july 2002.
- This was my first journal paper.

There are a number of my earlier conference works that are not listed here.

Research and study material

3D Reconstruction from a Handheld camera: a project I'm coordinating as a TA.
Summary of multiple view geometry as in Hartley's book (pdf | ppt)
Summary of a recent paper in Auto-Calibration (pdf | odp)
Obsolete Work Home Page

Software

VxL C++ Computer vision libraries. This is what we use here at Brown.
SIP: Image processing toolbox for Scilab, similar to Matlab but with absolute freedom
Hotreference.com: build your own searchable bibliography database online, with lots of storage space; Bibtex/doc export. Discuss you favorite paper with other users. All free (the site is by a very good friend of mine)
Gentoo Linux: an excellent OS for computer scientists and experienced users.

Photo album

Ask me by email for a username and password to access my album HERE

Email

Links

My brother's sonic compositions

Renato Fabbri

My wife's websites

Guz Design
From Both Sides - a graphic design thesis on hispanic immigration issues.

Beautiful Sea Shells

www.shellworldflkeys.com

Some Comics and Animations by my Cousin and Friends

Pula Pirata Website

No to Microsoft's OOXML/DOCX

www.noooxml.org

Home Pages of Other PhD Students at LEMS

Curiosities

Do the vertical borders of this page look strange to you? They are exactly the same gradients, reflected vertically. On their own, they should look convex or concave. But your brain is hypothesizing a global light source, and both bars can't be both convex and consistent with the same light source. There are also biases in direction - the brain tends to favor certain directions for the light source over others. As you might know, this is the convex concave crater illusion, and is related to the hollow-face illusion.