One of the emerging technologies that have showed a great advance in the last years and still has a great potential for development is Biometrics, which tries to identify people by their physical characteristics (voice, facial features, fingerprint …). This project is focused, as well says its title, in face recognition. Automated face recognition aims to identify people in images or videos using pattern recognition techniques and is widely used in applications ranging from social media to advanced authentication systems. Whilst techniques for face recognition are well established, the automatic recognition of faces captured by digital cameras in unconstrained, real‐world environment is still very challenging, since it involves important variations in both acquisition conditions as well as in facial expressions and in pose changes. Nowadays, the commercial software packages most commonly used in face recognition utilize the Principal Component Analysis (PCA) technique or any derivation from it. Furthermore, the 99 % of computational packages used in investigation projects about facial recognition that use other statistic methods, share something in common with PCA: the Euclidean space. The software based on Euclidean metrics is not perfect recognising faces for two reasons: from one hand, due to the algorithm itself and from other hand, due to the input data to the system (faces). In this project are described the theory elements that build the Principal Component Analysis. Moreover, a computational system in MATLAB based in PCA was developed to analyse a face database with different photos per each person, pretending make a study about PCA technique modifying the training set, observing both weak points and strengths. An engineer should be able to solve the problems trying to provide the optimal solution, so another objective of the project is to explore alternative solutions in each step of the work, identifying the best one for each case.
