PERFORMANCE ANALYSIS OF FACE RECOGNITION SYSTEMS BASED ON FALSE REJECTION OR FALSE ACCEPTANCE OF PROBE IMAGE

Abstract

Face recognition like other biometrics systems involves some basic processes, which includes biometric feature acquisition / enrollment which in this case would be faces of human to be recognized, normalization of these enrolled features in order to standardize the training set and lastly is the recognition which involves mapping the enrolled features collected to features of people to be recognized i.e the probe images. Several comparisons have been made on some face recognition systems with variations in each of the result, even when the same algorithms is used in those experiments. This variation has in no small measure rubbish the authenticity of these algorithms leading to the common problem of either false acceptance or false rejection on the target object. This paper presents a comparative analysis of the performance of some selected face recognition systems, namely the PCA, 2DPCA and ICA. The algorithms were implemented and tested exhaustively to evaluate the performance of these algorithms under different face databases and similarity distance metrics in respect to the recognition accuracy. We statistically present the results obtained.