Design of Novel Approach for Conformal Vortex Induced Flow Based Shape Descriptor for Object Recognition

Abstract

The object complex plane is one of the important techniques for determination of system stability. In this plane, the vortex based flows are ubiquitous and their analysis and understanding remain an active research area. There are different directional patterns for flow computation viz., row, column, mesh, triangular, outside the cylinder and outside the vortex. The present paper proposes a novel approach for shape descriptor based on vortex outside the cylinder flow. The flow of velocity is represented with the equipotential lines and streamlines. The Fourier transformation is used for feature vector generation and Chebyshev distance measure is used for classification. The proposed descriptor is experimented on various standard databases viz., MPEG-CE-1, Kimia 99 and Kimia 216. The evaluated results are compared with the benchmark shape descriptors. The efficacy of the proposed descriptor is proved with the enhanced results than the benchmark descriptors.