Unsafe driving behavior of bus drivers is one of the main causes of passenger traffic accidents. In order to regulate the driving behavior, a detection method of steering wheel operation behavior of bus drivers based on hand feature extraction is proposed. Firstly, the Adaboost target hand classifier is constructed by multi-feature cascading detection method, and the driver's target hand detection is carried out in the region of interest of the steering wheel, in order to obtain the number and position information of target hand. Then, the position of the steering wheel in the detection image is calibrated to identify the position relationship between the target hand and the steering wheel. Finally, the classification rules for unsafe driving operations of bus drivers are formulated. The experimental results show that the average detection accuracy is 94.0%, the average detection speed is 30ms/frame. The method has good real-time performance and robustness.

CICTP