A person is able to recognize the faces of other people thanks to the brain area on the border of the occipital and temporal lobes - the fusiform gyrus. People learn to recognize faces from birth and already at four months old can clearly distinguish one person from another. The main things a person pays attention to are the eyes, cheekbones, nose, mouth and eyebrows, as well as the texture and color of the skin. At the same time, our brain processes the face as a whole and is able to identify a person even by half of the face. The brain compares the resulting picture with the internal average template and finds characteristic differences. Therefore, it seems to people that representatives of another race are "all in one face": the internal templates of people are tuned to facial features characteristic of their environment. How does the information system of face recognition work - more on that in today's issue!
First of all, the face recognition system needs to find a face in the image and select this area. To do this, the software can use a variety of algorithms: for example, determining the similarity of proportions and skin color, highlighting the contours in the image and matching them with the contours of faces, highlighting symmetries using neural networks. The Viola-Jones method, which can be used in real time, is considered the most effective. With it, the system recognizes faces even when turned 30 degrees. The method is based on Haar features, which are a set of black and white rectangular masks of different shapes. Masks are superimposed on different parts of the image, and the algorithm adds up the brightness of all pixels in the image that are under the black and white parts of the mask, and then calculates the difference between these values. Then the system compares the results with the accumulated data and, having identified the face in the image, continues to track it to select the optimal angle and image quality. For this, motion vector prediction algorithms or correlation algorithms are used.
Having selected the most successful pictures, the system proceeds to face recognition and its comparison with the existing database. It works on the same principles as the artist draws portraits - the program finds anchor points on a person's face, from which individual features are formed. As a rule, the program allocates about 100 such points. The most important measurements for face recognition programs are the distance between the eyes, the width of the nostrils, the length of the nose, the height and shape of the cheekbones, the width of the chin, the height of the forehead, and other parameters.
When using 2D images, it is possible to successfully recognize a face only when shooting from the front view and in good lighting, which is suitable for security systems in enterprises and government agencies. For work in public places, 3D images are used. Several synchronized cameras take a number of photographs from different angles, on the basis of which a three-dimensional model of the object is formed, with which the system works, determining control points. After that, the obtained data are compared with those available in the database, and, if the parameters match, the person is identified.
In addition to 3D models, scientists are developing other areas. For example, Identix has created a highly accurate biometric facial recognition technology that analyzes skin texture - pores, lines and scars. According to the developers, the use of their technology together with the traditional face recognition system will increase the accuracy of work by 25%.
In the next installment we will talk about how the banknote detector works. Stay with us!
Promotional video: