Comparison of eigenvector-based pattern recognition algorithms for hybrid systems

Pattern recognition algorithms based on eigenvalue analysis for hybrid processing (optical-digital computer) are theoretically and experimentally compared. These algorithms consist of the Foley-Sammon (F-S) transform, Hotelling trace criterion (HTC), Fukunaga-Koontz (F-K) transform, linear discriminant function (LDF), and generalized matched filter (GMF). It is shown that all these algorithms can be represented in a generalized eigenvector form, and that they differ in the ways in which they utilize the correlation matrices (F-K) or covariance matrices (F-S, HTC, etc.) to calculate the discriminant vectors. Some methods classify the images, or, instead, features of the images, in a reduced dimension. The above algorithms are tested experimentally by using 20 training images and 10 test images, all with 64*64 pixels.<>