Motif discovery from 2D arrays and applications
- 14:00 25th November 2014 ( week 7, Michaelmas Term 2014 )Lecture Theatre B
The analysis of digital images often involves the search of rectangular boxes that are frequently repeated in one or several input images. Digital images may be modeled by 2D arrays, from which motif patterns consisting of sequences of intermixed solid and don’t-care characters are extracted. In order to alleviate the exponential growth of such motifs, notions of maximal saturation and irredundancy have been formulated, whereby more or less compact subsets of the set of all motifs can be extracted, that are capable of expressing all others by suitable combinations. In this talk, the notion of maximal irredundant motifs in two-dimensional arrays will be explained and a combinatorial argument that poses a linear bound on the total number of such motifs will be presented as well. Then different approaches to the discovery of 2D irredundant motifs will be discussed and some applications for digital images compression and classification will be illustrated.