How it works? | Face Detection
Man is able to recognize the faces of other people through the brain area on the border of the occipital and temporal lobes - fusiform gyrus. People are learning to recognize faces from birth and in four months, can clearly distinguish one person from another. The main thing that people pay attention - this is the eyes, the cheekbones, nose, mouth and eyebrows, as well as texture and color. In this case, our brain processes the person as a whole and is able to identify a person even half of the face. The brain compares the resulting image with the inner averaged pattern and finds characteristic differences. Therefore, it seems to people that members of other races "all look the same": the internal patterns of people attuned to facial features, specific to their environment. How does the information face recognition system - about it in today's issue!
First of all, the face recognition system is necessary to find the person in the image and select the area. To do this, the software can use a variety of algorithms, for example, determining the similarity of proportion and color, edge detection on the image and compare them with the contours of persons, the allocation of symmetry using neural networks. The most effective is the Viola-Jones method that can be used in real-time. With it the system recognizes the faces even when rotated 30 degrees. The method is based on the Haar signs, which represent a set of black and white rectangular masks of various shapes. Masks are applied on various parts of the image, and the algorithm adds the brightnesses of all pixels of the image caught by black and white portions of the mask, and then calculates the difference between these values. The system then compares the results with historical data and identify the person in the image, it continues to monitor for the optimal angle and image quality. For this purpose, the motion vector prediction algorithms or correlation algorithms.
Selecting the most successful pictures, the system starts to face recognition and comparison with the existing base. It works on the same principles as the artist paints portraits - the program finds the reference points on the human face, of which the individual features are added. Typically, a program 100 is allocated about such points. The most important measurements for facial recognition programs are the distance between the eyes, width of the nostrils, nasal length, height and shape of the cheekbones, the width of the chin, forehead height and other parameters.
When using 2D-images successfully identify a person can only be shot from the front and with good lighting that is suitable for security systems in enterprises and public institutions. To work in public places used 3D-image. Several synchronized cameras do a number of pictures from different angles, which are formed on the basis of a three-dimensional model of the object, with which the system works, identifying the control points. After that, the data are compared with a database, and for matching parameters, the person is identified.
addition of 3D-models, scientists are developing other directions. For example, the company has created a high-precision Identix biometric face recognition technology that analyzes the texture of the skin - pores, lines and scars. According to the developers, the use of technology together with the traditional face recognition system will increase accuracy by 25%. In the next issue we will talk about how to operate the banknote detector. Stay with us!