Object Recognition Using Symbolic Similarity Analysis | Original Article
Object recognition is a processof extracting information such as size, position, pose and functions related to the object. Var-ious approaches have been developed using invariant, 3D-CAD models (K. Roh and I. Kweon 1998, M. Stevens and J. Beveridge, 1997. D. Keren,M. Osadchy and C. Gotsman, 2001) .Object recognition methods based on local image such as local differential in-variants and SIFT are developed but these approaches have shown limited success to some problems (David.G.Lowe.1999, C. Schmid,A. Zesserman and R. Mohr, 1998). In this paper object recognition system is developed using Zernike moments, and symbolic simi-larity analysis. In this paperZernike moments are used,as these moments are invariant to general affine transformations. Here the concept of symbolic object and similarity measure are used for better object recognition under rotation and scale changes. The system consists of two parts, in first part is training phase and second part is testing phase. In training phase the object from COIL-100 database are trained and values are stored in the form of symbolic object in MATLAB database. In testing phase a query image is given as input for which a symbolic object is created using Zernike moments and recognition of the query image is carried out by doing similarity analysisbetween test object and knowledge base. The system is developed on MATLAB 7.6.0 and run on Pen-tium machine.