Automatic Recognition Of Catenary Rotating Binaural Defects Based On OpenCV

As the core component of electrification railway overhead catenary,the state of catenary support and suspension devices directly determines the stability of whole suspension system,effecting the catenary security status,in the event of a failure,it will affect the safety of railway traffic.Rotating binaural as an important part of catenary support and suspension devices,play the role of connecting the positioning tube and the oblique wrist arm.Its once the failure will cause the positioning tube and the locator to fall,tilt or sag,resulting in a serious catenary accident.Currently,Within the popularization and application in railway catenary suspension state detection monitoring device is mainly focused on the insulator,carrier cable detection and so on,while the rotating binaural defect state recognition is still in the manual interpretation state,the efficiency is low,the degree of automation is limited.For rotating binaural defect recognition has not been applied in actual,this article embarked on the catenary rotating binaural defect recognition algorithms.In this paper,the catenary state detection and monitoring device(4C)in the actual line of images collected as the object of study,analyzed and studied the application of catenary geometry parameters,insulator fault identification,positioning clamp fault identification and support vector machine(SVM)combining feature extraction methods in face recognition and other fields,this paper presents a method for identifying and detecting the rotating binaural of the catenary based on image processing and recognition.And more than one actual circuit test data proves the feasibility of this method.Firstly,the quality of the detected image is improved by image preprocessing,and acquire the high quality picture.Image preprocessing mainly includes image denoising,contrast enhancement and global binarization.And then through the method of feature extraction and support vector machine positioning the rotation of the binaural region,extracting the cotter pin of rotating binaural region,and perform a fault detection of the cotter pin missing.In the positioning of the rotating binaural region,this paper applied the LBP feature and the HOG feature combined the support vector machine respectively.And put forward a mix LBP-HOG features of LBP features and HOG to rotating binaural region identification,Implementing the rotating binaural region location identification.In the rotating binaural region defect detection,locate No.1 cotter pin through the Gabor feature and support vector machine.And on this basis,based on the distance transformation method to accurately locate the No.2 cotter pin.In the cotter pin region which is precisely positioned,the state of the cotter pin is judged by the method of identifying the end of the cotter pin,and achieve the pin missing fault identification in rotating binaural region cotter pin.Finally,this paper verified the feasibility and reliability of rotating binaural intelligent recognition algorithm,by testing rotating binaural images of catenary suspension state detecting and monitoring device(4C)under different working conditions.