Object recognition based on topology preserving skeleton features

Abstract

Object recognition is a procedure for recognising a particular object in an advanced video or image. Appearance-based or feature-based techniques are used for object recognition. An object skeleton is the useful cue for object recognition, which provides a structural representation to specify the relationship among object parts. The shape's geometry and topology can be efficiently encoded. In this paper, an effective object recognition method is introduced with the help of multi kernel SVM using the skeleton derived from topology preserving skeleton features method.