The Research And Application Of Face Recognition In Video Retrival

The thesis has made a series of researching on face recognition-related problems, including automatic face detection, feature extraction, face recognition. As for color image in video, some novel methods are proposed in feature extraction and face recognition. The main researching includes the following:1, In face detection, we first use Adaboost method, which finds the number of faces and their approximate positions in the image. After transforming color images to gray images, combining with eye location method based on hybrid projection function, the accurate position of eyes can be located. The experiment results demonstrated that the method is more robust to poses, illumination, etc, especially suitable for detecting rotated faces in image plane.2, The thesis has made some researching on feature extraction and face recognition based on subspace methods. Gabor transform is also researched for feature extraction to get more effective features which is more robust to poses, illumination. Combining with Fisher method to extraction the key points, we just do Gabor transform on these points to get real-time performance.3, A method is implemented. It extract Gabor feature, and then SVM classifies these features. Firstly train the example images based on Fisher method which can get a subspace .Then use this subspace to get key points in an image on which is done Gabor filters on to get Gabor features. At last, these features are classified by SVM. The method performs very well in our experiments.