How to Realize Dynamic Face Recognition

There are many methods of Face recognition. The main methods of Face recognition are:

(1) Face recognition Based on 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 the advantages of fast recognition speed, small memory requirement, but low recognition rate.

(2) Face recognition method based on feature face (PCA): feature face method is a face recognition method based on KL transform, and KL transform is an optimal orthogonal transform for image compression. After KL transformation, a new set of orthogonal bases is obtained in high-dimensional image space.
The important orthogonal bases are preserved and can be extended into low-dimensional linear spaces. If we assume that the projections of faces in these low-dimensional linear spaces are separable, we can use these projections as feature vectors for recognition, which is the basis of feature face method.
This idea. These methods need more training samples and are based on the statistical characteristics of image gray level. At present, there are some improved feature face methods.

(3) 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. This method also needs more samples for training.
In many applications, the number of samples is very limited.

(4) Face Recognition Method Based on Elastic Map Matching: Elastic Map Matching defines a distance that is invariant to normal face deformation in two-dimensional space, and uses attribute topology to represent face, any of the topological graphs.
Each vertex contains a feature vector, which is used to record face information near the vertex position. This method combines the gray level characteristics and geometric factors, and expounds the recognition principle, application scenarios and future development of artificial intelligence branch-face recognition system.

In comparison, the elastic deformation of the image can be allowed, which can overcome the influence of expression change on recognition. At the same time, the training of multiple samples is no longer needed for a single person.

(5) Line Hausdorff Distance (LHD) Face Recognition Method: Psychological research shows that human beings are no worse than recognizing gray scale image in speed and accuracy of contour recognition (such as cartoons). LHD is based on extracting gray-scale images of human faces
It defines the distance between two sets of line segments. What is different is that LHD does not establish a one-to-one correspondence between different sets of line segments, so it is more adaptable to minor changes in line segments. The experimental results show that:
LHD performs very well under different illumination conditions and different postures, but its recognition effect is not good in the case of large expressions.

(6) Face Recognition Method of Support Vector Machine (SVM): In recent years, support vector machine is a new hotspot in the field of statistical pattern recognition. It tries to make the learning machine compromise on experience risk and generalization ability, so as to improve the performance of the learning machine.
Support Vector Machine (SVM) mainly solves a two-classification problem. Its basic idea is to transform a low-dimensional linear inseparable problem into a high-dimensional linear separable problem. The usual experimental results show that SVM has a good recognition rate.
However, it requires a large number of training samples (300 in each category), which is often unrealistic in practical applications. Moreover, the training time of support vector machine is long and the implementation of the method is complex. There is no unified theory for the selection of the function.