A new object recognition framework based on PCA, LDA, and K-NN

In this paper, a new object recognition framework is presented. The framework includes a variety of object recognition approaches based on Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and the K-nearest neighbor (K-NN). A color image vector representation model is also introduced. Based on the representation model, color Eigenspace is constructed using PCA and LDA for feature extraction. The K-NN is adopted as a mechanism for classification. The object recognition framework approaches were tested on Columbia Object Image Library (COIL-100) and Amsterdam Library of Object Images (ALOI) color object databases. Experimental results on these databases evinced the effectiveness of the proposed work with high recognition rates in comparison with the results of previous works in the literature. The recognition rate has reached in some cases 100%.