W I L L K O M M E N
	Universität Freiburg - Institut für Informatik Lehrstuhl für Mustererkennung und Bildverarbeitung (LMB)

LMB Mitarbeiter Forschung Studien-, Diplom- und Bachelor... Algorithmen zur Klassifikation... Trifocal Tensor Estimation () Evaluation of Blind Deconvolut... Matching of crystal surfaces Lernen von visuellen Objektkla... Comparison of Polar Fourier Tr... Subpixelgenaue Bewegungsschaet... Konturdetektion in zeitaufgelo... Bildverarbeitungspraktikum I Project: Bildrektifikation Project: Active Countour und S... Lehrveranstaltungen Veranstaltungen Publikationen Awards () Stellenangebote TILDA Software () Interna  Diese Seite optimiert für Ausdruck [No english page available]

ALBERT-LUDWIGS-UNIVERSITÄT FREIBURG INSTITUTE FOR COMPUTER SCIENCE Chair of Pattern Recognition and Image Processing Prof. Dr.-Ing. Hans Burkhardt Georges-Köhler-Allee 52, Room 01-029, D-79110 Freiburg, Tel. 0761-203-8260 The Chair of Pattern Recognition and Image Processing is currently offering a Studienarbeit (Student research project) on Evaluation of Blind Deconvolution Algorithms Blind deconvolution refers to a class of problems of the form: $$g(x) = h(x)*f(x)+n(x)$$ (1) where $I \subseteq R^2$ is the support of the image, and $f(x), g(x), h(x)$ and $n(x)$ represent, respectively, the unknown original image, the observed image, the unknown impulse response or point spread function (PSF) of the blurring system, and the observation noise. The operator ($*$) in (1) denotes 2-D-convolution, given by $$(h*f)(x)=\sum_{s\in S}h(s)f(x-s)$$ (2) where $S \subseteq R^2$ is the support of the PSF. The goal of this work is to: Report about the state-of-the-art blind deconvolution methods Implement blind deconvolution algorithms ([1], [2], [3]) on toy data sets Test the algorithms on microscopy images Bibliography 1 F. Sroubek and J. Flusser, 'Multichannel blind iterative image restoration', IEEE Transactions on Image Processing 12(9): 1094-1106 (2003). 2 J. Markham and J. Conchello, 'Parametric blind deconvolution of microscopic images: further results', Proc. SPIE 3261:38-49, (1998). 3 P.J. Verveer, J. Swoger, F. Pampaloni, K. Greger, M. Marcello and E.H.K. Stelzer, 'High-Resolution tree-dimensional imaging of large specimens with light sheet-based microscopy', Nature Methods 4(4): 311-313, (2007). Candidates: Students in computer science, physics, mathematics or other related fields. Good knowledge of C++ is required. Please contact: Maja Temerinac-Ott Room: 01-0460 Telephon: 0761/203-8275 E-Mail: temerina@informatik.uni-freiburg.de Jun. 2008 Figure: Deconvolution in 2D: One image was distorted by three unknown PSFs. After blind deconvolution, the original image as well as the three PSF functions are estimated. [three distorted images of the same object] [3 estimated PSF functions and the restored image] Figure: Deconvolution in 3D: The original form of the sphere (red) should be recovered from the recorded sphere convolved with the PSF function. The goal is to estimate the PSF function. [xy-view] [xz-view] [yz-view] [3D-view]