Identification Of Typical Violations Of Coal Mine Safety Production Based On Video Analysis

Violations in coal mine safety production are one of the reasons for coal accidents.Therefore,violation detection based on video surveillance is the core technology to achieve safe coal production.At present,the mine video image analysis technology is not yet mature,and there are little researches on the identification of mine violations.To finish off this problem,taking mine video surveillance as the research object,the following three innovative types of research on underground video are carried out:(1)A video surveillance image enhancement algorithm for complex mine environments is proposed.According to the guide filter,the lighting component is extracted and the image brightness value is set by the two-dimensional gamma function.and the CLAHE is used to adaptively enhance the image contrast,to realize the adjustment of the uneven illumination of the underground video image and the serious phenomenon of dust at the same time.The experiment compared 3 different scenes.The results show that the ameliorated algorithm has a significant improvement in visual effects over the conventional Gamma function and the Retinex algorithm and by comparing information entropy,average gradient,and standard deviation,the improved algorithm has significant advantages.(2)Propose an improved algorithm for the detection and tracking of underground personnel.A combination of background subtraction and Kalman filtering could deal with the automatic people moving underground detecting and tracking.Directing at the matter of false detection caused by flashlight shaking,it is raised to preclude detection disturbance based on pixel brightness.The video data of the flashlight shaking randomly in different scenes is selected,and the detection results show that the improved algorithm can filter out the miner's lamp area in the video image,and is contrasted with the traditional algorithm in detection rate,false-positive rate,and false-negative rate.Improved,the detection rate of the video images of two examine different scenarios is increased by 15.38% and 11.11% respectively.It shows that the algorithm has the characteristics of excellent precision and strong robustness.(3)Designed a model for identifying violation behaviors underground.For underground personnel,there may be two typical violations of taking off their helmets and breaking into dangerous areas.The color and shape characteristics of the helmet are extracted to match the video image,and the position of the helmet is judged according to the human body structure ratio to identify the behavior of removing the helmet;the spatial topology relationship is used to determine whether it is in a dangerous and sensitive area.The results show that these two behaviors modeling models are suitable for judging illegal behaviors of taking off the safety helmet and intruding into dangerous areas underground,and have strong stability and high practicability.The identification of typical violations based on video analysis has a strong application in coal safety production.It solves the shortcomings of manual all-weather monitoring of underground video images to9 achieve the use of intelligence.The use of a video monitoring and analysis system to prevent accidents is of great meaning to coal safety production.Figure 34;Table 5;Reference 78...