Image recognition using Artificial Immune systems approach

Abstract

This paper describes the initial stage in developing the concepts of Artificial Immune System (AIS) in solving engineering problems such as pattern recognition and optimization. The purpose of this paper is to perform an analysis on the pattern recognition using AIS approach. The negative selection algorithm (NSA) has been selected as a tool to solve the problem due to its simplicity as compared to other immune models and algorithms and its suitability to model pattern recognition problem. Binary matching rules are usually implemented in NSA since binary strings provide easy manipulation in computer and easy to reason with and display. This paper compares the performance of two different binary matching rules; the Hamming distance matching and the r-contiguous bit matching rule in distinguishing the non-self pattern from the self pattern in pattern recognition problem. The results obtained show the percentage rate of detection accuracy for both matching rules. It can be concluded that both matching rules provide high detection rate if the threshold parameter value is decreased. Finally, conclusions of the study are presented and future direction work is specified.