Working as postdoc at FEF-Unicamp, since 04/04. Current work involves Camera calibration, tridimensional reconstruction and pose recovery.


Nice to see what I wrote about my PhD at its start. Funny how things tend to go sideways while not completely going off topic. Anyway, PhD done. I'm doing the rounds on all websites and updating the informations. I'll post the details of my PhD here soon, source code included. If, by any chance, you, dear reader, need it and its not  here yet, feel free to contact me.


I'm a student of Université Paris-Est now, working at the ESIEE, under supervision of Prof. Laurent Najman. My project aims to create new morphological filters to enhance images, in a general way, but  focusing OCR pre-processing.


Short (5 pages) version of master's thesis:



One picture of the UPA 2008 demo robot. My father helped me with the mechanical construction of the robot, I did the eletronic part (motors/uController interfacing USB) and the computational part (it had 2 webcams for computer vision based control).

It's just a demo, so it can be controlled via wifi or  ordered to keep itself towards the green disc.


This "post" is not about my current research. Only programming is never enough, you have to really participate in lab work:


 It's been a while since my last "report" here. Fortunately I have some very good news :) . My aim is to investigate the use of 3D information applied to tracking of coletive sports.

  • The method for background identification (for sports) is working perfectly. I'm writing an article about it, should be a short paper, but I maybe have enough work for a full paper... 
  • I've tested the main reconstruction method in sintetical videos, and it works really fine. But the calibration problem is still there... at least is easily solved - just a better calibration pattern. And it works without the necessity of computing matching points for the reconstruction.
  • The method for the real tracking of the players is almost done and it is incredible fast! (at least for me...). Furthermore, the ant colony variation I've created is interesting.
  • Another byproduct of the work is the generalisation of the concept of visual rhythm - widely used to detect video cuts - used to tracking of targets. This piece will be in an article just after the background identification and the report to FAPESP.

I'll put some overviews presentations here soon, showing this work. If everthing works as intended, I'll achieve a really interesting degree of automation.

Interested? mail me : /


I'm translating some parts of this site to english. Well, my english is not that good, and some errors may appear. Please notify me of them:

My current research project is about tridimensional tracking of handball players, aiming their kinematics analysis. My advisors are:


Proposta de mestrado quase pronta! O Ricardo Barros revisa as coisas muito rápido! Falta bem pouco para acabar a proposta agora.


Minha proposta de mestrado que vou encaminhar pra fapesp está sob revisão dos meus orientadores... não quero nem ver o que eles vão falar!

A proposta é de melhoria das ferramentas desenvolvidas no LIB pelo Pascual e companhia, para rastreamento tridimensional de jogadores de handebol.  Vai ser legal, com várias câmeras, reconstruções tridimensionais, muitos dados, demorado pra processar, etc etc etc, tudo do jeito que eu gosto!

Vou trabalhar sob orientação do prof. Neucimar do Instituto de Computação, meu orientador desde 2003 e sob co-orientação do prof Ricardo Barros do Laboratório de instrumentação para a biomecânica da FEF.

Adicionei também o link para minha única publicação até o momento, o relatório técnico da minha iniciação.