Research On Detecting The Locomotive Drivers' Fatigue Based On Computer Vision

Security is the most important in railway traffic. The locomotive drivers'fatigue is one of the key causes for the railway accidents. Thus, a system for exact, real-time detecting of the locomotive drivers'fatigue is needed to improve the security of locomotive driving.This thesis mainly focuses on two vital issues. One is how to develop an algorithm based on computer vision for detecting locomotive drivers'fatigue. The other is how to establish a system for detecting locomotive drivers'fatigue according to the present algorithm and the practical requirement. The main contents of this thesis are as follows. Firstly, the behavior mode of locomotive driving is analyzed. The principle of causing locomotive drivers'fatigue is studyed by consulting the survey of locomotive drivers'fatigue. After analyzing and concluding the existence methods for detecting the drivers'fatigue, the noncontact method based on passive computer vision is selected. According to the selected method, the main scheme of locomotive drivers'fatigue detecting system is proposed.Secondly, an algorithm for detecting the locomotive drivers'fatigue based on PERCLOS is proposed. The algorithm includes three steps, i.e. face detection, eye location and PERCLOS calculated after fatigue feature extraction. For face detection, skin area is segmented using the assembling character of skin color in YCbCr color space. Then, the face is enhanced using the morphology. After that, human face structure features are used to locate the face. For eye location, a new method using integral projection based on mask is proposed. The method reflects the edge and the gray of the pixels in a particular direction of the images. It can exclude the disturbance from forelock, the sideburns, the collapsible, and the hat etc by using mask. At last, the PERCLOS is adopted to recognize the drivers'fatigue. The experiments show that the algorithm is effective and reliable.Lastly, the system of locomotive drivers'fatigue detecting is built with TMS230DM 6437 as the core processor. The detecting system consists of the image collection module, the image processing module, the expanded memory module, the voice warning module and the electrical source manage module. The image collection module is used to collect the face images of locomotive driver, and also to convert analog images into the digital images. Then, the algorithm of detecting the locomotive drivers'fatigue is accomplished by the image processing module. The results of fatigue detection are put out by RS485. If locomotive drivers'fatigue is determined, the voice warning module can be controlled whether to alarm or not. Thus, the security of locomotive drving is ensured by the system.