Research on feature extraction of mechanical operators with abnormal behavior

Abstract

High frequency of safety accidents in the process of mechanical production has brought great pressure and loss to the social economy. Relevant statistics show that the proportion of illegal operation of operators is the main factor. In order to reduce the occurrence of safety accidents, the computer vision research technology based on monitoring video is applied to the direction of mechanical safety behavior recognition. The content of abnormal behavior of mechanical operators is relatively broad. A foundation for the subsequent simulation analysis can be formed through extracting the characteristics of the actor, calculating the characteristic operator of human behavior, and debugging and modifying the threshold value of the characteristic operator for judging the abnormal behavior.