This paper compares various face recognition techniques to address the challenges posed by partial occlusion, particularly in the eye and bottom face regions. The study finds that using Spatially Confined Non-Negative Matrix Factorization (SFNMF) yields the highest recognition rates, achieving 95.17% for bottom face recognition and 96.7% for full faces. The results highlight the effectiveness of partial face information in scenarios such as access control where complete face visibility is not feasible.