Method based on geometric features
Face recognition using geometric features usually uses the location of important feature points such as eyes, mouth and nose and the geometry of eyes and other important organs as classification features. However, roder did experimental research on the accuracy of geometric feature extraction, and the results were not optimistic.
Local feature analysis method
Locality and topology are ideal characteristics for pattern analysis and segmentation, which seems to be more in line with the mechanism of neural information processing. Therefore, it is very important to find expressions with such characteristics. This method has achieved good results in practical application. It forms the basis of faceit face recognition software.
Feature face method
Feature face method has simple and effective points, which is also called face recognition method based on principal component analysis. From the statistical point of view, the basic elements of face image distribution, that is, the eigenvector of covariance matrix of face image sample set, is found to approximately represent human face image. These feature vectors are called feature faces.
Method based on elastic model
Elastic image matching technology is a recognition algorithm based on geometric features and wavelet texture analysis of gray distribution information. Because the algorithm makes good use of the structure and gray distribution information of human face, and has the function of automatic and accurate positioning of facial feature points, it has a good recognition effect.
Neural network method
Lee et al. Described the characteristics of human face with six rules, and then located the facial features according to these six rules, and input the geometric distance between the five facial features into fuzzy neural network for recognition. Compared with the above methods, the application of neural network in face recognition has certain advantages, and it is more adaptive and generally easy to realize.