The Eye Behavior Recognition Algorithm Analysis Based On Computer Vision

Computer vision is a hot research topic in recent years, it also is a comprehensive discipline. Computer vision refers to using the camera and computer replace the human eye and brain recognition of target locating and tracking processing means such as technology subject usually. Eyes are key parts of the body and eyes positioning, detection, identification, analysis and processing to face recognition, iris recognition, medical test, expression analysis and so on related research to provide the necessary technical support and abundant reference information.As image processing, pattern recognition, data mining and the development of the computer information processing technology, more and more algorithm that based on computer vision technology research and application is have emerged. In this paper, on the basis of the existing technical research and the ideas of classical algorithm, we improved algorithm and puts forward a series of new algorithm, Through the experiment proves the rationality of the new algorithm feasibility.Firstly this paper proposes a fusion texture and geometric features of eye detection algorithm, Achieve the eye location detection in image.In order to analyze the dynamic state of video streaming eyes behavior, tracking algorithm has been introduced in this paper. This paper introduced a new algorithm, and has made the improvement of STC algorithm, achieved the dual target tracking. Experimental results show the improved algorithm is fast and reliable.It is a key link part that eye image processing analysis in the process of human eye condition analysis, this paper puts forward a new window-filtering algorithm. and than used a new algorithm for regional segmentation in order to further study.In the next part we put forward a simple and rapid static images of the human eye state classification recognition algorithm. In order to solve the problem, an improved simplifying method was put forward, classical support vector machine classifier realized the image of the human eye open and close state identification.Finally this paper proposes a statistical algorithm of blink, defines the pupil sensitivity parameters, eye fatigue and eye strain state are analyzed, The experimental results meet expectations.