Morphological Boundary Based Shape Representation Schemes on Moment Invariants for Classification of Textures

MORPHOLOGICAL BOUNDARY BASED SHAPE REPRESENTATION SCHEMES ON MOMENT INVARIANTS FOR CLASSIFICATION OF TEXTURES V. Venkata Krishna1 and M. Rama Bai2 1C.I.E.T, JNTUK, Kakinada, Rajahmundry, Andhra Pradesh, India, E-mail: [email protected] 2Dept. of CSE, M.G.I.T, JNTUH, Hyderabad, Andhra Pradesh, India, E-mail: [email protected] Efficient shape representation and recognition is an important issue in image processing and computer vision. It provides the foundation for the development of efficient algorithms for many shape related processing tasks. One of the disadvantages of these shape representation schemes is that they yield a poor classification and recognition rate. The classification requires a human intervention, thus the shape representation and classification methods are not automatic. To address these problems the present paper presents a novel and effective methods of shape representation by morphological boundary based methods. The shape features are evaluated by the proposed morphological boundary based methods by suitable numerical characterization derived from moment invariant measures for a precise classification. The proposed Morphological Boundary based Shape Representation scheme (MBSR) derives a new shape descriptor to address the image classification problem by combining boundary extraction and Hu moment (HM) invariants information. The proposed novel schemes of shape representation are applied on original, noisy, rotated and scaled images. The experimental results clearly show the efficacy of the present method.