The Design And Implement Of Work Scene Security Detection System Based On Computer Vision

Most of the current substations are unattended.For some unsafe factors in the substation,the monitoring personnel of the remote monitoring center are completely staring at them.This will inevitably lead to leakage due to human negligence,and after the failure,on-site maintenance People often spend a lot of time and effort on inspections to find fault points.Various types of warning signs are hoisted in the substation scene to warn people to avoid danger;fire-fighting equipment placed in a fixed position can be fired in time in the event of fire;insulators play an important role in electrical insulation and mechanical fixing of high-voltage equipment.effect.The fall of warning signs,the removal of fire-fighting equipment,and the damage of insulators can hinder the safe operation of substations.Therefore,we need a set of work scene safety analysis system to intelligently monitor these common safety hazards in the substation,detect faults and make early warnings,make up for the possible negligence caused b y manual monitoring,and reduce the effective positioning of fault points.The inspection and positioning work of the maintenance personnel saves a lot of time and energy.This paper designs a machine vision-based work scene safety analysis system,which includes monitoring of the falling off of warning signs,monitoring of fire equipment removal,and monitoring of insulator damage.In the warning sign detachment and fire equipment removal monitoring,according to the unique color and shape characteristics of the warning signs and fire-fighting equipment in the substation scene,the color-based image segmentation and shape-based ROI extraction recognition and positioning methods are proposed.The precise positioning of the warning sign and the fire-fighting equipment,based on the positioning information,monitors the positioning part to detect whether the warning sign in the area has fallen off and whether the fire-fighting equipment has been removed.In the insulator damage monitoring,an insulator identification and localization method based on HOG feature and SVM classifier is proposed.For the problem that the SVM classifier is too dependent on the number of samples,the image is segmented and morphologically processed,and then sent to the SVM for classification.Detection,in order to realize the identification and positioning of the insulator;based on the positioning information,the ellipsoidal arc of the insulator is corrected by affine transformation.Finally,according to the unique shape feature o f the insulator,the rule of the grayscale statistical strip is used as the characteristic criterion.Its damage identification.Finally,through the function and performance test of the system,the feasibility of the warning sign off monitoring,the fire equipment removal monitoring and the insulator damage monitoring algorithm and the reliability of the system are verified.