Research On Human Fatigue State Detection Method Based On Machine Vision

With the development of modern society,the pace of life is accelerating gradually.It is difficult to ensure adequate rest for many people,and it is easy to cause fatigue during work.Fatigue has a bad influence on the safety of work production especially for driving,monitoring and dispatching.It is of great significance for the safety of production to detect the workers who are in the fatigue state in time during the working process.The research of human fatigue detection technology has become the focus of the worldMachine vision inspection technology is a non-contact automatic detection technology.Its major principle is to use the camera instead of the human eye to obtain image information,and process the image,so that the machine can realize detection,judgment,recognition,measurement and other functions.In this paper,the main methods of human fatigue state detection in the world are classified and summarized.According to the requirement of human fatigue state detection,the human fatigue state detection by machine vision technology is researched.The main research work are as follows:1.The principle of human body fatigue state detection method based on machine vision and the main visual fatigue characteristics are introduced.2.The Adaboost face detection algorithm is deeply studied.Based on the VS2010 platform,the face detection program is written,and the face area is detected by using the face cascade classifier in Open CV visual library.3.On the basis of face detection,the rectangular region of face is divided.According to the "three-courts and five-eyes" rule of human face and other influencing factors in the process of actual fatigue detection,the candidate region of human eyes are divided.The human eyes classifier is used to locate the human eyes accurately in the candidate region of human eyes.The rectangular region of the human eyes is redivided according to the accurate position of the human eye.Combined with the P-tile method,the rectangular region of human eye position is binarized,the human eyes segmentation is realized,and the human eyes contour is extracted.4.This paper collect the face videos of different states and carry out experiments.The actual fatigue state and PERCLOS value of the measured object were calculated and the appropriate PERCLOS threshold was set.And the human fatigue state detection based on PERCLOS is realized.