Realization Of Fast Spring Bar State Recognition Based On Texture Features

With the development of transportation,railway transportation has become one of the main modes of transportation in China.Rail is an important part of the railway,through the rail train to transfer its own pressure to the sleeper,so that the body can run smoothly and smoothly on the line.Once the rail is bent,broken or deviated,it will cause the train to shake,bump and even derail.Because the connection between rail and sleeper is realized by track fastener,track fastener can keep the continuity and integrity of rail,prevent the rail from moving relative to the sleeper,and act as a shock absorber system when the train is running,so the status identification of track fastener is correct It is of great significance to improve the safety of railway operation.As a kind of bolt-free track fastener,Pande Road fast elastic strip fastener(SFC,Slab Fast Clip)has the characteristics of large buckle pressure,less parts and good ride comfort,and is gradually replacing the traditional fastener to be used in railway.considering that the traditional track fastener state recognition algorithm can not effectively identify the state of the SFC fast elastic strip,this paper studies the fastener state recognition algorithm for the SFC fast elastic strip.First,the obtained SFC fast projectile image is preprocessed.Through histogram equalization,the influence of noise on image quality is reduced,and the sharpness and contrast of picture are improved.Yes After that,the fuzzy clustering algorithm is used to process the image,and the contrast degree of different regions of the image is further improved,thus reducing the difficulty of SFC fast projectile bar positioning.and then,the cross-cross method is used to locate the SFC rapid projectile strip to remove the interference information in the image.on the basis of this,we use the ACE algorithm to enhance the texture of the elastic strip,weaken the texture of the background,and use the gray-gradient co-occurrence matrix to extract the texture features of SFC fast elastic strip.meanwhile,Gabor filter is used to enhance the high-frequency components of the image of the fastener area and to extract the shape features of SFC fast elastic strip.Then we use parallel fusion to treat two species features for fusion.finally,the fused features are put into the SVM for training,and the trained SVM classifier is used to identify the state of the SFC fast elastic strip.The algorithm presented in this paper is verified by using 500 images of track fastener collected by linear array camera as experimental material.Experimental results show that this algorithm can effectively solve the problem of SFC fast bar state recognition,and has good accuracy,false detection rate and missed detection rate.