Federico Pernici Phd   

 currently post-doc, University of Florence.

  Address: Via di Santa Marta No. 3, 50139 Firenze - Italy  
  Phone:  +39 55 4796540
  Web: http://viplab.dsi.unifi.it/~pernici/

Doctoral Dissertation
Two Results in Computer Vision using Projective Geometry

Federico Pernici  PHD Thesis

University of Florence Faculty of Engineering Dipartimento di Sistemi e Informatica (2006)

Selected Publications


"Towards on-line saccade planning for high-resolution image sensing" 

Alberto Del Bimbo, Federico Pernici 

Pattern Recognition Letters, In Press, Corrected Proof, Available online 2 May 2006, 


"Distant targets identification as an on-line dynamic vehicle routing problem using an active-zooming camera"

Alberto Del Bimbo, Federico Pernici

Visual Surveillance and Performance Evaluation of Tracking and Surveillance, 2005. 2nd Joint IEEE International Workshop on 15-16 Oct. 2005 Page(s):97 - 104 

"Accurate automatic localization of surfaces of revolution for self-calibration and metric reconstruction". 

Carlo Colombo, Dario Comanducci, Alberto Del Bimbo, Federico Pernici

In Proceedings IEEE Workshop on Perceptual Organization in Computer Vision POCV 2004, Washington, DC, USA, June 2004, IEEE 2004. 

"Metric 3D reconstruction and texture acquisition of surfaces of revolution from a single uncalibrated view"

Carlo Colombo, Alberto Del Bimbo, Federico Pernici

IEEE Transactions on Pattern Analysis and Machine Intelligence
vol. 27, no. 1, pp. 99-114, Jan. 2005

Uncalibrated 3D metric reconstruction and flattened texture acquisition from a single view of a surface of revolution.

Carlo Colombo, Alberto Del Bimbo, Federico Pernici

1st International Symposium on 3D Data Processing Visualization and Transmission (3DPVT'02)  June 19 - 21, 2002 Padova, Italy

"Image Mosaicing from Uncalibrated Views of a Surface of Revolution". 

Carlo Colombo, Alberto Del Bimbo, Federico Pernici

In Proceedings British Machine Vision Conference (BMVC), 2004 Kingston University, UK Sept. 7-9.    PRESENTATION(.ppt)


Single View Reconstruction : VRML 3D Models 








  •  Click on the thumbnails to download.

  • These are zipped files containing the textured 3d models reconstructed from a single image (also included).




                                                                              Click the images for the 3d Model (vrml).

Past Projects

Current Projects


In this project we propose a method based on uncalibrated video analysis techniques to correctly compute the observation model from broadcast sport video sequences acquired by active zooming cameras. In these sequences the observation model changes frame by frame. For example the 3D players position measured in frames with high zoom, is more accurate (less uncertain) than the position measured in wide angle images. The importance of the correct propagation of such uncertainty is very important when the camera moves and players exit and enter from the field of iew. In this condition keeping players identity at later time in the sequence is very challenging especially for the scale change due to zooming effect. The problem will be even more complicated when the appearance of targets does not discriminate between targets. We argued that good performance in prediction (which is strongly dependent on the observation model) can improve the robustness of target tracking especially during disappearance and reappearance of target objects.


Research Interests