What Are the Methods of Face Recognition_ Taigewang Technology

First, the face recognition method of geometric features: geometric features can be the shape of eyes, nose, mouth and the geometric relationship between them (such as the distance between them). These algorithms have high recognition speed and small memory, but the recognition rate is low. Second, face recognition method based on feature face (PCA): feature face method is a face recognition method based on KL transform, which is an optimal orthogonal transform of image compression. After KL transformation, a new set of orthogonal bases is obtained from the high-dimensional image space, and the important orthogonal bases are retained, which can be expanded into a low-dimensional linear space. Assuming that the projections of human face in these low dimensional linear spaces are separable, these projections can be used as feature vectors for recognition, which is the basic idea of feature face method. These methods need more training samples, and are completely based on the statistical characteristics of image gray. At present, there are some improved feature face methods. Third, face recognition method of neural network: the input of neural network can be face image with reduced resolution, autocorrelation function of local region, second-order moment of local texture, etc. Such methods also need more samples for training, and in many applications, the number of samples is very limited. Fourth, the face recognition method of elastic graph matching: the elastic graph matching method defines a distance that is invariant to the usual face deformation in two-dimensional space, and uses the attribute topology to represent the face. Any vertex of the topology contains a feature vector to record the information of the face near the vertex. This method combines the gray characteristics and geometric factors, allows the image to have elastic deformation during comparison, and has achieved good results in overcoming the influence of expression change on recognition. At the same time, it does not need multiple samples for training for a single person.