Investigating the Effect of Illumination and Viewpoint on Image Recognisability

View point and illumination are two main factors that make difficult for machines to compare the two instances of the same object. Therefore investigating the effect of illumination and viewpoint on images recognisability is crucial. How ever the goal of our final project is to build a system which can achieve understanding from images. Thus this paper is focusing on investigating how an artificial system can separate the images based on view point and illumination. The technique we used in this experiment is image clustering. The growing self organizing map was used as the clustering tool as it has many advantages over many other clustering tools as well as the ability to generate hierarchical clustering. The experiments were carried out using a well known face image database called 'Yale face database B'. The results of this experiment showed the potential of using growing self organizing map to separate objects with different effects into different groups.