The Research on Detection and Recognition of Driving Posture

In order to effectively understand and characterize driving behaviours, this paper proposes an efficient feature extraction approach for driving postures by combining test image preprocessing methods and skin-color regions segmentation. The technology consists of reference white, Laplace algorithm, skin-color model, edge detection and connected regions detection, SVM classifier, etc., which effectively realizes the detection and classification of four driving postures including grasping the steering wheel, operating the shift lever, eating a cake and playing on the phone. The test accuracy of 250 samples from a driving posture dataset reaches 97%.