Driver smoking behavior real-time monitoring method based on vehicle video monitoring

Abstract

The invention discloses a driver smoking behavior real-time monitoring method based on vehicle video monitoring. The method includes the following steps of acquiring vehicle monitoring camera video data; extracting the luminance component of monitoring video data; setting an interested area and performing thresholding on the interested area; after the thresholding, sequentially performing morphological corrosion calculation, morphological dilation calculation, and the template-based region growth on the setting luminance region in the video frame to obtain an initial mask image; subtracting the video frame after the morphological corrosion calculation from the video frame after the morphological dilation operation to obtain a mask image; obtaining a characteristic luminance value after the thresholding; and determining whether there has smoking behaviors based on the amount of the nonzero pixels in the characteristic luminance value after the thresholding. According to the method, the real-time monitoring and alarming for driver smoking behaviors on large passenger vehicles can be realized without artificial participation.

Description

technical field [0001] The invention relates to a method for real-time monitoring of drivers' smoking behavior based on vehicle-mounted video monitoring. Background technique [0002] Large passenger vehicles are an important means of transportation. In recent years, with the continuous increase of such vehicles, the problems caused by traffic have become increasingly prominent. Among them, bad driving behavior of drivers is the main cause of traffic accidents. These driving behaviors include mobile phone calls, smoking, and fatigue driving while driving. Among the bad behaviors of the above-mentioned drivers, smoking behavior may lead to the following adverse consequences: distracting the driver's attention and causing accidents; possibly causing fires in the car; seriously deteriorating the air quality in the car, causing discomfort to passengers, and causing complaints from passengers. Effective remote monitoring and real-time early warning of drivers' smoking behavior...

AI Technical Summary

Problems solved by technology

[0003] The current detection methods for drivers' smoking behavior include: using multi-information fusion methods to detect smoking behavior; using information such as vehicle speed, acceleration and deceleration to prompt the driver; using color space to detect the driver's hand, and then through machine learning Realize action recognition and use Gaussian function modeling to realize the detection of the driver's abnormal behavior mainly for fatigue driving; however, none of the above methods can specifically realize the real-time monitoring of the driver's smoking behavior

Images