A multiwavelet model for 2D object analysis and classification

A multiwavelet based model utilizing the GHM scaling and wavelet functions has been developed for object classification. Multiwavelets are employed as they possess such important properties as orthogonality, symmetry, and compact support, which cannot be achieved simultaneously with scalar wavelet analysis. Additionally, multiwavelets allow the use of two or more inputs concurrently. In the context of shape analysis this permits modeling both coordinates of an object contour sequence. All these properties make multiwavelets an idea tool for object analysis and classification. The proposed algorithm is invariant to change in scaling, shifting, rotation, and starting point. The scaling function expansion coefficients, which contain maximum energy, are used as features for object classification. Experimental results demonstrate the model's efficacy.