Algorithms for human face recognition

1, the content of human face recognition technology

Human face recognition technology contains three parts:

(1) human face detection

Face detection refers to the dynamic scene with a complex background to determine whether there is a face, and to separate such a face. There are generally the following methods:

① reference template method

first design one or several standard face templates, and then calculate the degree of match between the samples collected in the test and the standard template, and through the threshold to determine whether there is a face;

② face rule method

Since the face has a certain structural distribution characteristics, the so-called face rule method is to extract these features to generate the corresponding rules to determine whether the test sample contains a face;

③ Sample Learning Method

This method adopts the method of artificial neural networks in pattern recognition, i.e., to generate classifiers through the learning of face sample sets and non-face sample sets;

4 Skin Color Model Method

This method is based on the relative concentration of the distribution of face and skin color in the color space. color space to detect the pattern of relative concentration of distribution in the color space.

⑤ Feature sub-face method

This method considers the set of all face images as a sub-space of face images and determines whether there is a face image based on the distance between the detected samples and their projections in the sub-holes.

It is worth proposing that the above five methods can also be synthesized in practical inspection systems.

(2) Human Face Tracking

Face tracking refers to dynamic target tracking of detected faces. Specifically, model-based methods or methods based on a combination of motion and modeling are used.

In addition, the use of skin color model tracking is not a simple and effective means.

(3) Human Face Matching

Face matching is to identify the detected face image or to search for the target in the face image database. This effectively means that the sampled face images are compared with the stock face images in sequence and the best match is identified. Therefore, the description of the face image determines the specific method and performance of face recognition. At present, two description methods are mainly used:

①Feature vector method

The method is to determine the size, position, distance and other attributes of the five facial features such as the iris of the eye, the nose, and the corners of the mouth, and then calculate their geometric eigenquantities, and these eigenquantities form a feature vector that describes the facial features.

② Facial pattern template method

This method is to store a number of standard facial templates or facial organ templates in the library, and in the comparison, all pixels of the sampled facial image are matched with all templates in the library using normalized correlation metrics.

In addition, there are autocorrelation networks for pattern recognition or a combination of features and templates.

The core of human face recognition technology is actually "local human characterization" and "graphical/neural recognition algorithms." This algorithm is the use of human facial organs and characteristic parts of the method. Such as the corresponding geometric relationship between multiple data formation recognition parameters and all the original parameters in the database for comparison, judgment and confirmation. The general requirement of judgment time is less than 1 second.

2, the human face of the recognition process

Generally divided into three steps:

(1) First of all, the establishment of the human face of the face file. That is, with a camera to collect unit personnel of the human face of the face file or take their photos to form the face file, and these face files to generate the face pattern (Faceprint) code stored.

(2) Get the current human face image

That is, the camera captures the current access to the face of the personnel, or take the photo input, and the current face image file to generate the faceprint code.

(3) with the current face of the code and archive inventory comparison

That is, the current face of the face of the code and archive inventory of the face of the code to retrieve the comparison. The above mentioned "Face Code" method works on the basis of the essential features and beginnings of the human face. This facial coding is resistant to changes in light, skin tone, facial hair, hairstyle, eyewear, expression, and posture, and is highly reliable, making it possible to accurately recognize a person from millions of people.

The process of recognizing human faces can be done automatically, continuously, and in real time using common image processing equipment.