A robust intelligent face recognition framework using GNP-based multi-agent system

Previously, a principal component analysis (PCA) based face recognition framework using Genetic Network Programming (GNP) and Fuzzy Data Mining (GNP-PCA) was proposed to improve both the accuracy and robustness of the conventional PCA-based face recognition algorithm in the complicated illumination database. However, it is still not robust enough in the noisy testing environments. Therefore, a GNP-based multi-agent system is constructed by GNP-PCA and multi-resolution analysis in this paper. In the proposed approach, the different scales of images in the training set are regarded as different environments and each GNP-PCA is performed as an agent in each environment. Recognition is eventually realized by evaluating the prediction scores for different classes. According to the experimental results, the proposed method has almost no accuracy loss in the Gaussian noisy testing environments compared with GNP-PCA.