A novel alternative to conventional machine vision features. (27th January 2009)

• Record Type: Journal Article
• Authors: Deb, Suash
• Is Part Of: International journal of bio-inspired computation. Volume 1:Number 1/2(2009)
• Page Start: 89
• Page End: 92
• Publication Date: 2009-01-27
• Subjects: global features -- local shape descriptors -- extended features -- extended ratios -- extended angles -- bio-inspired computation -- machine vision -- feature selection -- polygons -- polygon reconstruction; Biologically-inspired computing -- Periodicals; Computational biology -- Periodicals; 572.0285
• Languages: English
• ISSNs: 1758-0366

Abstract

The problem of recognition, classification and distance measurement between different polygons are all very important tasks in pattern recognition, text identification and model-based vision. The performance of any such job depends to a great extent on the proper selection of features. Towards that, this paper introduces a new set of shape descriptors – extended angle and extended ratio. These arise out of every set of alternating line segments of any polygon. These are invariant under similarity transformation (Deb, 2006). These together with a proper measure of the direction of the extended lengths makes the feature set unique in the sense that besides recognition, it helps in completing the reconstruction of the unknown polygon. This paper will study several characteristics of the extended feature set.