Research On Track Damage Recognition Algorithm Based On Image Edge Detection

With the rapid development of our country’s railway construction,the length of railway transportation has increased,the operation cycle has increased,and the transportation volume and efficiency have been greatly improved.Heavy-duty and high-speed railways have become the main infrastructure to promote my country’s social and economic development.The railway track is in all-weather operation.In order to improve the safety of railway operation,real-time observation of its status is required.Therefore,higher requirements are put forward for the identification of damage to the existing railway track structure.Image acquisition and information extraction have the advantages of contactlessness,intelligence and high speed.Due to the development of software and hardware technology,image processing technology has been improved.Applying it to the field of damage identification of existing railway tracks is the direction of future development.With the support of hardware systems and algorithms,this paper improves the image recognition algorithm,combines edge information detection and feature extraction with track structure damage recognition,studies the track structure damage adaptive recognition algorithm,and realize the damage identification of rail surface,sleeper,fastener and other areas.The specific work done is as follows:(1)Using matrix reset and determinant calculation methods,study the smoothing algorithm of orbit image based on improved two-dimensional convolution.The algorithm first supplements the edge elements of the image pixel matrix,resets them according to the matrix row and column elements,makes full use of the edge pixels,and combines the traditional two-dimensional convolution with the "left row and right column" transformation calculation method of the unit matrix.Propose a two-dimensional convolutional image smoothing algorithm that is modified to improve image quality and increase processing speed.(2)According to the light and dark characteristics of the rail image and the special shape skeleton of the fastener spring bar,the rail surface and the fastener image are separated.The improved two-dimensional convolution image smoothing method and the bimodal threshold setting are applied to the traditional Canny edge detection algorithm,and compared with the traditional edge detection algorithm,the edge detection result of the track image is obtained.The Hough transform is used to extract the edge line type of the track surface,the included angle is calculated according to the slope of the straight line,and the angle and length information existing in the edge line type of the track image are extracted to further realize the damage identification of the sleeper and the rail surface.(3)Research on fastener damage recognition algorithm based on image feature point matching,extract image features(color,texture,geometric features,etc.)into the process of generating feature point descriptors.Construct the feature point descriptors of the image of the fastener area,and by sorting and sampling them,the edge feature point data in the image collection is matched in different images according to the original template,and the image matching is completed.According to the correct matching rate of feature points,analyze and compare the fracture and missing phenomena of fasteners.And use the image texture to compare the features of the matched images to verify the accuracy of the judgment and recognition results.