| Popis: |
In surveillance scenarios, identifying a person captured on image or video is one of the key tasks. This implies matching faces on every still photos and video sequences. A Automatic face recognition for still photos with high quality will do satisfactory performance, aside from video-based face recognition it's exhausting to achieve similar levels of performance. Compared to still pictures face recognition, there are several disadvantages of video sequences. First, pictures captured by CCTV cameras are typically of poor quality. The background level is higher, and pictures could also be blurred because of movement or the topic being out of focus. Second, image resolution is generally lower for video sequences. If the topic is incredibly far away from the camera, the actual face image resolution can be as low as 64 by 64 pixels. Last, face image variations, like illumination, expression, pose, occlusion, and motion, are additional serious in video sequences. The approach can address the unbalanced distributions between still images and videos in a robust way by generating multiple abridges to connect the still images and video frames. So, in this project, we can implement still to video matching approach to detect the face from videos using Grassmann manifold learning approach and also recognize the faces using neural network algorithm to know unknown matches. Finally provide SMS alert at the time unknown matching in real time attendance environments. |
