What is face recognition

Face recognition specifically refers to the use of computer technology to analyze and compare the visual characteristics of human face information for identification. Face recognition is a popular field of computer technology research, it belongs to the biometric identification technology, is the organism (generally refers to people) itself to distinguish between the biological characteristics of the individual organism. The broad face recognition actually includes a series of related technologies to build a face recognition system, including face image acquisition, face localization, face recognition preprocessing, identity confirmation and identity search, etc.; and the narrow face recognition refers to the technology or system of identity confirmation or identity search through the face. Biometric features studied by biometrics include face, fingerprints, palm prints, palm type, iris, retina, vein, voice (speech), body shape, infrared temperature spectrum, ear type, odor, personal habits (such as the strength and frequency of keyboard, signature, gait), etc., and the corresponding recognition technology has face recognition, fingerprint recognition, palm print recognition, iris recognition, retina recognition, vein recognition, voice recognition (with voice recognition can be used to identify the face). Recognition (with voice recognition can be identity recognition, but also the recognition of voice content, only the former belongs to the biometric identification technology), body shape recognition, keystroke recognition, signature recognition and so on. Face Recognition Methods with Geometric Features Geometric features can be the shape of the eyes, nose, mouth, etc. and the geometric relationship between them (e.g., distance from each other). These algorithms are fast in recognition and require little memory, but the recognition rate is low. Face Recognition Methods Based on Feature Face (PCA) The feature face method is a face recognition method based on the KL transform, which is an optimal orthogonal transform for image compression. The high dimensional image space is transformed by KL transform to get a new set of orthogonal bases, retaining the important orthogonal bases in it, from which it can be converted into low dimensional linear space. If the projections of the face in these low-dimensional linear spaces are assumed to be differentiable, these projections can be used as feature vectors for recognition, which is the basic idea of feature face methods. These methods require a large number of training samples and are based entirely on the statistical properties of the image grayscale. There are some improved feature face methods available. Face Recognition Methods with Neural Networks The input to a neural network can be a face image with reduced resolution, an autocorrelation function of a local region, second order moments of a local texture, and so on. This type of method also requires a larger number of samples for training and in many applications the number of samples is very limited. FACE RECOGNITION METHODS FOR ELASTIC GRAPH MATCHING The elastic graph matching method defines a distance in a two-dimensional space with some invariance for usual face deformations, and uses an attribute topology graph to represent the face, where any vertex of the topology graph contains a feature vector to record the information of the face in the vicinity of the vertex position. The method combines grayscale properties and geometric factors to allow for elastic deformation of the image during comparison, and receives better results in overcoming the effects of expression changes on recognition, while multiple samples are no longer required for training for a single person. Line Hausdorff Distance (LHD) for Face Recognition Psychological studies have shown that the speed and accuracy of human beings in recognizing contour maps (e.g., caricatures) is no worse than that of recognizing grayscale maps.LHD is based on line segment maps extracted from grayscale images of faces, which defines the distances between two sets of segments, and, uniquely, does not establish a one-to-one correspondence of the segments between the sets of segments, so that the distance between the segments can be calculated as a one-to-one correspondence. of one-to-one correspondence between different line segment sets, so it is more adaptable to small changes between line segment maps. Experimental results show that LHD has excellent performance under different lighting conditions and different postures, but it is not good at recognizing large expressions. Support Vector Machines (SVM) for Face Recognition In recent years, Support Vector Machines are a new hot spot in the field of Statistical Pattern Recognition, which tries to make the learning machine reach a compromise between the empirical risk and the generalization ability, so as to improve the performance of the learning machine. Support vector machines mainly solve a 2-classification problem, and its basic idea is to try to transform a low-dimensional linearly indivisible problem into a high-dimensional linearly divisible problem. Usual experimental results show that SVM has a good recognition rate, but it requires a large number of training samples (300 per class), which is often unrealistic in practical applications. Moreover, the support vector machine training time is long, the method is complicated to implement, and there is no unified theory for the kernel function to be taken. New Face Recognition Technology Traditional face recognition technology is mainly based on visible light image face recognition, which is also the most familiar recognition method, and has more than 30 years of research and development history. However, this approach has insurmountable defects, especially when the environmental lighting changes, the recognition effect will be sharply reduced, and can not meet the needs of the actual system. The solution to the problem of illumination has three-dimensional image face recognition, and thermal imaging face recognition. However, these two technologies are still far from mature, and the recognition effect is not satisfactory. A recently rapidly developing solution is the active near-infrared image-based multi-light source face recognition technology. It can overcome the effects of light variations and has achieved excellent recognition performance, with overall system performance exceeding that of 3D image face recognition in terms of accuracy, stability and speed. This technology has developed rapidly in the last two or three years, making face recognition technology gradually move towards practicality. Digital camera face autofocus and smiley face shutter technology The first is face capture. It determines according to the part of the human head, first determines the head, then judges the head features such as eyes and mouth, and confirms that it is the human face through the comparison of the feature library to complete the facial capture. Then the human face is used as the focal point for autofocus, which can greatly improve the clarity of the photos taken. Smile shutter technology is based on face recognition, complete facial capture, and then begin to determine the degree of upward curvature of the mouth and the degree of downward curvature of the eyes to determine whether it is smiling. All of the above capture and comparison is done in the context of comparing feature libraries, so the feature library is the foundation, which contains a variety of typical facial and smiley face feature data.