Research On Workers’ Fatigue Detection Method Based On Video Processing

Currently, it has become a serious problem caused by workers with tiredness,and it can’t be ignored. Fatigue can make people work slowly, attention decreaseand so on. Under the special circumstances at the scene of the electric power,fatigue of the workers will lead to be negligent, not alert, wrong operation andcause serious consequences. Fatigue is a hidden security incident, not be takenseriously, and there is no uniform standard, so the fatigue detection technology isdifficult to apply into real life. This article studies the fatigue detection methodof workers, including video images processing, face detection, eye and mouthfeature extraction and method of fatigue detection, mainly analysis the featureextraction methods of eyes and mouth.In the study of the human eye detection method, this article has presented amethod of combining YCb’Cr’ skin model and Otsu algorithm to extract color andnon-color region. Firstly, in YCb’Cr’ color space, images are segmented intodifferent regions with skin-color information to narrow the eye search area. Then,the segmented image is converted to a binary image by Otsu algorithm to extractthe eye regions. Finally, the left and right eyes are positioned in the facial areawith the binary integral projection.In the study of the mouth contour feature extraction method, this articleadopt SUSAN algorithm to solve this problem. Firstly, mouth contour edge isextracted in the segmented face image by SUSAN algorithm. Then, opening andclosing operation in mathematical morphology are used in the image to removenon-mouth region. Finally, a feasible method for detecting missing parts of themouth is proposed by searching the optimal region around the region of mouth,improved the detection rate of mouth contour extraction.This article uses Matlab、 C++language to achieve the simulationexperiment on the localization of eyes and mouth, then analysis the data offatigue status. The experimental result reveals that this algorithm has achieved thedesired goals, verified the validity and rationality.