Comparison of Colour Image Discriminant and Fisher Linear Discriminant Algorithms

Abstract: Recognition by face is one of the biometric methods used commonly for identification. The Colour Image Discriminant (CID) model tries to combine the colour image exemplification and recognition tasks into one framework. Basic CID colour space, extracts multiple features in the colour image, where three new colour component images D1, D2 and D3 are derived using an iterative algorithm.The classical FLD method involves only one set of variables: one or multiple discriminant projection basis vectors, for image discrimination. Experimental results using two face databases, namely, Face96 and Indian Face dataset of IIT Kanpur and then with live still images succeeds in gender recognition. The method achieves the recognition rate of 88.43% with ROC of 0.903 showing the effectiveness of CID algorithm.

Keywords: Colour component images, colour image discriminant (CID) colour space, gender recognition, Fisher linear discriminant analysis (FLD or LDA).