Eye Control System Based On OpenCv And MFC Of PC Side

With the rapid development of science and technology and people’s living standards improvement, brand new human-computer interaction technique has attracted more and more attention. Eye-control technology has liberated man’s hands. Vigorous development of this new technology will significantly enrich the normal entertainment. This article focuses on the pupil center detection method for the low-resolution picture, Haar classifier, image morphological processing and exponential weighting filtering method. On this basis, I developed an eye control system for low-resolution images. This research contains the face and eye detection, eye parameters calculating and time series filtering, UI design and systems analysis.In real time face and eyes detection, we located the face position using Haar classifier in OpenCv Toolbox after comparing multiple face detection algorithms and tools, then narrowed the scope to identify eye position by position of facial features and proportions. This method is real-time, insensitive to ambient light. Besides, user can sit at ease when using this software. We found that if the user is sitting in front of a screen at around 60 cm, face detection rate can reach 99.99%, eye detection rate is 92%, detection time is about 39 to 45 milliseconds per frame.In the process of calculating eye parameters, we put forward two parameters: pupil vector and eye height. After image graying and thresholding, floodfill algorithm was used to find the maximum connected region where the iris and pupil locate. This method has many advantages, such as real-time computation, high accuracy. What’s more, it can reduce the effect of blinking and strong adaptability to different individuals. In real-time parameter filtering, exponential weighting filtering method was used to average time sequence data. This method sorted newest data by displacement, removed quarter of maximum and minimum values, finally returned the weighted summation of left three-quarter values which base number is one-second sorted by time order. Finally we use this method for its high robustness and cut the adverse effects of head movement.In UI design, MFC interface can show pupil’s center position, four keypad location, eye image, important parameters, system settings and other information in real time, which help user easily master the status of eye control system. In system analysis, several games are designed to test system stability, accuracy and efficiency.Experimental result shows on the condition that light is suitable and user keep upright sitting, this eye control system can press a button in one second, the horizontal accuracy is up to 90%, vertical accuracy is above 70%, comprehensive accuracy is about 80%. Due to the low frequency and high accuracy of push-button, the eye control system can meet practical requirements.