Image Detection Method for Machine Element and Tread Damage of Heavy Duty Goods Wagon
The present detection of machine element and tread damage of heavy haul goods wagon is mainly implemented manually,which is affected by the area and angle of the parts and the subjectivity of the manual,and whose detection accuracy needs to be improved.This paper presents an image detection method for damage of machine element and tread of heavy-duty goods wagon.The images of machine element and tread of heavy haul goods wagon are collected,and homomorphic filtering method is applied to eliminate noise in the image and enhance image contrast ratio.With Canny edge detection algorithm,the edge of the collected image is detected,and the damage area of Machine element and tread is segmented to obtain the target area.The VGG-19 network is established on the basis of transfer learning,and the above target areas are input into the VGG-19 network to gain the damage characteristics of machine element and tread,so as to realize the damage detection of machine element and tread.The experimental results show that the image processing effect of the proposed method is favourable,the detection accuracy of mechanical parts damage and tread damage is higher than 90%and the detection time is less than 10 s,which is practical.
transfer learningheavy haul railway freight carsmachine element and tread damage detectionhomomorphic filteringCanny edge detection algorithm
国家科技期刊平台
NSTL
万方数据
