Research On CCD Surface Image Reconstruction And Defect Feature Extraction Of Outer Surface Of Inner Ring Of Train Bearing

An important reason for driving safety is the vehicle accident caused by the failure of the bearing.Therefore,the rolling bearing is a crucial part of the railway train transportation safety.Therefore,the railway department attaches great importance to the inspection and maintenance of the train bearing.By manually observing the rolling surface of the inner ring in the gap between the roller and the cage,the damage degree and defect type of the bearing inner ring and the roller surface are judged by hand feeling,visual inspection and people’s experience.Therefore,to solve these problems and meet the needs of production,a more advanced method is introduced to detect surface defects.This paper explores several common bearing surface defects(scratch,bump,crack,pitting,burn),image pretreatment,image Mosaic,image segmentation algorithm.The main research work and research contents are as follows:(1)Image acquisition of defects on outer surface of bearing inner ring.Machine vision detection is a process of image acquisition and image processing and analysis.Therefore,the quality of the collected images will directly affect the accuracy and stability of the machine vision detection.For image acquisition,this paper selects LED light source through the comparison of various light sources on the market,and USES CCD camera to collect the image of the outer surface of the bearing inner ring.(2)Image preprocessing.In the process of collection and transmission of images for a variety of factors,makes the image often appear all sorts of noise,the noise will make the image will not be able to fully reflect the real scene,so the need of image preprocessing,this article through to all kinds of filter experiment contrast,finally choose the suitable denoising effect better filter,this paper in order to gain more satisfactory image.(3)Image Mosaic processing.Due to the small perspective of the endoscope,a defect cannot be fully displayed,which makes it more difficult to identify the defect and reduces the accuracy.In this paper,a cylindrical projection surface correction method is proposed for the bearing’s surface property,then feature points are extracted and images are matched and fused,and a satisfactory Mosaic image is obtained.(4)Image segmentation and geometric feature extraction.After using different segmentation methods,the joining together of image segmentation,through analysis and comparison,selection of matching image segmentation based on threshold segmentation method,at the same time the use of mathematical morphology to eliminate the interference information of the image after segmentation,to the geometric feature extraction of image after segmentation,provides a basis for algorithm of recognition.using traditional after joining together with in this paper,the defects after stitching method of image segmentation and template image comparison,discovered in this paper,the splicing method of the defect image closer to the template image segmentation,the splicing method of this paper is more outstanding.