<![CDATA[Face recognition involves matching facial features by restricting the facial area. The problem found in the experiment was that the program recognized images outside the face area, especially for low-resolution images. The PCA algorithm and the proposed bounding box approach can identify the facial area and match it with training data. The experiment uses the Yaleface and Face94 datasets in various scenarios, including normal resolution and resolution reduction (75%, 50%, and 25% of the original size). On gif images, the proposed algorithm can produce similarities between the detected image and the input image in a resolution reduction of up to 50%. On jpg images, reducing resolution to 75% does not affect the performance of PCA. The proposed method can recognize faces with similarities in variations of pose and facial expression. The Euclidean value of the jpg image produces a better similarity value than the gif image.]]>
Copyrights © 2024
