Skip to Main Content
Patten recognition techniques are often an important component of intelligent systems to describe, classify and recognise the objects. Object recognition using linear vector quantization neural network which is trained using descriptors such as boundary and regional descriptors is presented in this paper. Out of the various descriptors available, a combination of these descriptors extracted from the input patterns is proposed for recognizing objects using a vector quantization neural network. The feature vector obtained from the object is the combination of a boundary descriptor, the signature or and a regional descriptor, the Zernike moment magnitudes of the image. The results of the testing phase are included at the end of this paper.
Date of Conference: 18-20 Dec. 2008