|
Main research projects:
Scene analysis and interpretation: The plenoptic function
The plenoptic function describes what an "eye" would see from any viewpoint in space and in any direction at any time. In this project, we are investigating new scene segmentation and intrepretation schemes by analysing all the multiview data as a single function. Applications include scene interpretation and image based rendering (see below for some interpolated viewpoints).
Dataset: Imperial College Electrical and Electronic Engineering Department lobby
Dataset: Dwarves, courtesy of the middlebury stereo vision page
Papers:
J. Berent, P. L. Dragotti, M. Brookes, "Adaptive Layer Extraction for Image Based Rendering," invited paper to special session on 3D, Proceedings of the IEEE International Workshop on Multimedia Signal Processing (MMSP'09), Rio de Janeiro, Brazil, October, 2009. [pdf] This paper received the Top 10% Award.
J. Berent, P.L. Dragotti, "Plenoptic Manifolds: Exploiting Structure and Coherence in Multiview Images," IEEE Signal Processing Magazine, vol. 24, no. 7, pp. 34-44, November 2007 [pdf]
J. Berent, P.L. Dragotti, "Unsupervised Extraction of Coherent Regions for Image Based Rendering," in British Machine Vision Conference (BMVC'07), Warwick, UK, September 10-13, 2007 [pdf]
J. Berent, P. L. Dragotti, "Segmentation of Epipolar-Plane Image Volumes with Occlusion and Disocclusion Competition," Proceedings of IEEE International Workshop on Multimedia Signal Processing (MMSP'06), Victoria, Canada, October 3-6, 2006 [pdf]
Image Fusion: Extended depth of focus for multi-channel microscopy images
Microscopy images suffer from small depth of focus. In this project, we investigate the use of wavelets for image fusion in order to create a single fully focused image from a stack of images taken at different focal lengths. Innovations include the use of complex wavelet transforms and a series of consistency checks that have proven to provide very good results.
I worked on this project while I was at the Biomedical Imaging Group (BIG) at the Swiss Federal Institute of Technology (EPFL). All the related information and the freely downloadable plugin are available on their website (http://bigwww.epfl.ch/demo/edf/).
Papers:
B. Forster, D. Van De Ville, J. Berent, D. Sage, M. Unser, "Complex Wavelets for Extended Depth-of-Field: A New Method for the Fusion of Multichannel Microscopy Images," Microscopy Research and Technique, vol. 65, no. 1-2, pp. 33-42, September 2004. [pdf]
B. Forster, D. Van De Ville, J. Berent, D. Sage, M. Unser, "Extended Depth-of-Focus for Multi-Channel Microscopy Images: A Complex Wavelet Approach," Proceedings of the Second 2004 IEEE International Symposium on Biomedical Imaging: From Nano to Macro (ISBI'04), Arlington VA, USA, April 15-18, 2004, pp. 660-663. [pdf]
Sampling theory: Piecewise Sinusoidal Signals Most digital acquisition devices involve the conversion of the observed signal from analog to digital. Classical sampling methods rely on bandlimitedness for perfect recovery of the original continuous time signal. In this project, we are investigating new sampling schemes for perfect recovery of other classes of signals. In particular, we are looking into parametric signals like piecewise sinusoidal signals.
Papers:
J. Berent, P. L. Dragotti, T. Blu, "Sampling Piecewise Sinusoidal Signals with Finite Rate of Innovation Methods," accepted for publication in IEEE Transactions on Signal Processing, July, 2009. [preprint]
J. Berent, P. L. Dragotti, "Perfect Reconstruction Schemes for Sampling Piecewise Sinusoidal Signals," Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP'06), Toulouse, France, May 14-19, 2006, pp. 377-380. [pdf]
Complex Wavelets - User Interfaces During an undergraduate project, I built an extensive matlab graphical user interface for comparing different complex and directional wavelet transforms in 1D and 2D.
|
