[IPOL announce] new article: Exemplar-based Texture Synthesis: the Efros-Leung Algorithm

announcements about the IPOL journal announce at list.ipol.im
Thu Oct 31 15:09:50 CET 2013


A new article is available in IPOL: http://www.ipol.im/pub/art/2013/59/

Exemplar-based Texture Synthesis: the Efros-Leung Algorithm
by Cecilia Aguerrebere, Yann Gousseau, Guillaume Tartavel
Image Processing On Line, vol. 2013, pp. 213--231.
http://www.ipol.im/pub/art/2013/59/

Abstract
Exemplar-based texture synthesis aims at creating, from an input sample, 
new texture images that are visually similar to the input, but are not 
plain copy of it. The Efros--Leung algorithm is one of the most 
celebrated approaches to this problem. It relies on a Markov assumption 
and generates new textures in a non-parametric way, directly sampling 
new values from the input sample. In this paper, we provide a detailed 
analysis and implementation of this algorithm. The code closely follows 
the algorithm description from the original paper. It also includes a 
PCA-based acceleration of the method, yielding results that are 
generally visually indistinguishable from the original results. To the 
best of our knowledge, this is the first publicly available 
implementation of this algorithm running in acceptable time. Even though 
numerous improvements have been proposed since this seminal work, we 
believe it is of interest to provide an easy way to test the initial 
approach from Efros and Leung. In particular, we provide the user with a 
graphical illustration of the innovation capacity of the algorithm. 
Experimentation often shows that the path between verbatim copy of the 
exemplar and garbage growing is somewhat narrow, and that in most 
favorable cases the algorithm produces new texture images by stitching 
together entire regions from the exemplar.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tools.ipol.im/mailman/archive/announce/attachments/20131031/14be6402/attachment.html>


More information about the announce mailing list