The Research Of Eye Location And Tracking Based On Haar-Like Characteristic

In modern society, fatigue driving has become one of reasons of traffic accidents, so using various methods to detect the driver's fatigue state has become a hot research area. Using computer vision technology to detect the driver's fatigue state is one of the hot spots in the study, Firstly, it can detect the driver's eye movement condition and trajectory by using computer vision technology, and then using the trajectories and eye state and characteristics of human fatigue can effectively judge the driver fatigue or not, in order to remind drivers or send an alarm.In the driver fatigue detection by computer vision technology, the driver's eye location and tracking problem is the core content of the study, how to accurately locate the eyes'position and track the eyes determines the reliability of fatigue detection. Focused on these problems, the eye movement video was done in the image preprocessing firstly, and then the positions of eyes were located in the preprocessed video image. Finally, the trajectory of eyes were tracked by using the Kalman filtering. The main works are as follows:1?Several popular eye location methods in computer vision were introduced, and the merits and demerits of these measures were analysed by theoretical analysis and experimental results. Finally, the AdaBoost algorithm based on Haar like rectangle features was chosen as the eye location method.2?The method of eyes tracking combined eye location and Kalman filter was proposed. These experiments showed this technique had well characteristic of correctness and real-time, and it met the human eye localization and tracking eye'trajectory.3?The method of improved the human eyes localization by using the predictability of Kalman was presented. The predictability of Kalman filtering algorithm was used to get the coordinate position of the human eye at the next moment, and the search range of the human eye location was reduced by using this coordinate position. Then it improved the eye location algorithm, the computation was cut down, the computing speed of eye localization was improved, and the real-time of tracking eyes was enhanced.