The recognition of objects in digital images is a key problem in computer vision. It has received the attention of many researchers in order to evaluate and improve the performance of descriptors but mainly the shape descriptor. In this talk, we will consider the persistent homology which is an algebraic tool to measure the topological features of shapes of high dimensional data. We use complexes to represent continuous spaces and specially the cubical complexes that are considered the basis of digital images. Using cubical complexes inside of triangulation reduce significantly the size of complexes. We represent this algebraic characterization as bare-codes that are a finite union of intervals and which are considered the shape descriptor. The benefits of this approach will be presented at the last section of this talk in the Arabic handwriting recognition using structural and syntactic pattern attributes.

This presentation is part of Contributed Presentation “CP6 - Contributed session 6”

Authors:

My Ismail Mamouni (CRMEF Rabat)

Keywords:

cubical homology, digital images, image reconstruction, persistent homology, topological data analysis