Clustering of Human Face images with different rotation angles

The aim of this paper is to classify the different human face images in specific clusters such that the different face images for any given person will be compiled in one cluster, whereas the different face images of other persons will be set in other clusters. For each person at least five different face images are taken by choosing different rotation angles. We proposed three algorithms, the first algorithm used to determine the adaptive neighborhood for each individual, the second one used to prepare the data of the studied population before the implementations, and the third is an automatic clustering algorithm. Our proposed algorithms were applied on different face images of many persons which were taken from the Iraqi population and the ORL database of many persons. We obtained a good classification results with small rate of error. This research is implemented by using Pentium (4) PC and MATLAB 6.5