QUALITATIVE CLASSIFICATION OF AN AUTONOMOUS DIFFERENTIAL EQUATION AND ITS APPLICATION TO IMAGE RECOGNITION SYSTEM

QUALITATIVE CLASSIFICATION IN IMAGE RECOGNITION SYSTEM

Abstract

In (face) image recognition system, a set of test images is compared to a set of training images for verification. A basic requirement for effective matching of these face images is that each image is structurally unique/distinct. This uniqueness is important in applications such as the issuance of national identification card and bank verification number. Standard algorithms and software exist on image recognition systems but there is still room for additional efforts. The present paper considers the problem of creating unique images for a hypothetical model of face image recognition system, such that the faces in the training data and the test data are simulated. The paper presents the classification of a first order autonomous ordinary differential equation having a nonlinear quintic polynomial part via the critical points of the equation. This is accomplished using a novel classification method, which may be called differential structure method, that was earlier developed by the author. The relevant qualitative properties are existence and uniqueness of solutions. It is shown that forty six (46) qualitative classes or unique images can be generated for matching in a simulated environment. A higher number of images can be generated by using a higher degree polynomial differential equation. Essentially, the results in the paper are also applicable to population modelling.