Thirteenth International Conference on Design Principles & Practices

Abstract

Building a frame model to determine the profile of human behavior is a difficult task for computer vision. A lot of programs are being created and constantly improved to make computer systems capable for correctly capture the dynamic objects. In the field of human face recognition systems such as Face ID (Apple) and Future Attribute Screening Technology (FAST) have been created. Still, modern computer vision systems still have difficulty in recognizing objects of reality. One of the most pressing problems: working with the movement, with the dynamics of the face, and with constantly changing mimicry. This problem, facing the computer vision, is about difficulty in identifying the same face in different lighting conditions, as well as in condition of changes the trajectory of its movement, and also if the angle (relative to the perceived device) or facial expressions change. This paper describes the approaches to solving this problem on the basis of light triangulation grids, which function in the environment of augmented reality. The construction of the facial frame is possible, using the properties of the natural chiaroscuro modeling of the face, formed as a result of the using various lighting model types. Owing to this approach, it is feasible to create a basic frame of the face for its subsequent recognition in dynamics. By forming such model in a computer vision system, a person can be identified under various conditions of external environment and with their facial expression changing.