Discriminative wavelet shape descriptors for recognition of 2-D patterns

In this paper, we present a set of wavelet moment invariants, together with a discriminative feature selection method, for the classification of seemi ngly similar objects with subtle differences. These invariant features are selected automatically based on the discrimination measures defined for the invariant features. Using a minimum-distance classifier, our wavelet momen t invariants achieved the highest classification rate for all four differen t sets tested, compared with Zernike's moment invariants and Li's moment in variants. For a test set consisting of 26 upper cased English letters, wave let moment invariants could obtain 100% classification rate when applied to 26 x 30 randomly generated noisy and scaled letters, whereas Zernike's mom ent invariants and Li's moment invariants obtained only 98.7 and 75.3%, res pectively. The theoretical and experimental analyses in this paper prove th at the proposed method has the ability to classify many types of image obje cts, and is particularly suitable for classifying seemingly similar objects with subtle differences. (C) 1999 Pattern Recognition Society. Published b y Elsevier Science Ltd. All rights reserved.