Research On Bearing Roller Defect Extraction Technology Based On Double Threshold

As the core component of the manufacturing industry,bearing is one of the strategic basic industries that China has developed.As a key component in the bearing structure,the bearing roller plays the role of supporting and reducing the rotational friction.Therefore,its quality determines the performance and life of the bearing.In the manufacturing process of bearing rollers,various surface defects may occur due to mechanical,environmental or human factors,which will cause the quality to deteriorate,affecting the stability and service life of the bearing.Once the defective unqualified bearing roller products enter the market,it not only has a bad impact on the reputation of the company,but also has major safety hazards.Therefore,it is very important to detect the presence or absence of defects on the surface of the bearing roller during the manufacturing process.In order to solve the above problems,this thesis proposes a defect extraction algorithm based on threshold segmentation based on actual projects.By adaptively setting the threshold,the algorithm completely separates the defect area of the bearing roller image,and uses the false defect filtering to further eliminate unnecessary areas to achieve the extraction of the defect target.(1)Aiming at the problem that the common threshold segmentation algorithm can not completely segment the defects,an adaptive double threshold defect segmentation algorithm is proposed.First,through the histogram analysis,some standard histogram data are selected;secondly,the standard function is obtained by fitting the partial data;then the difference value between the image histogram envelope function and the standard function is obtained,and the grayscale ratio is determined according to the defect.For the phenomenon of low or high,the first and last ones are chosen to make the local difference value and the maximum gray level as the preliminary defect threshold.At the same time,in order to reduce the threshold error caused by the fitting,in the defect histogram of the threshold range,the gray level corresponding to the local maximum is selected as the defect threshold of the precise positioning;finally,the image segmentation is realized.The experimental results show that the proposed algorithm can improve the integrity of the defect area and meet the real-time requirements.(2)Aiming at the problem of pseudo-defects in the segmentation defects,a pseudo-defect filtering algorithm with optimized features is proposed.Firstly,the algorithm combines isotropic diffusion and morphological processing to obtain the region with local background.Secondly,it considers the influence of Local background crossing and noise interference,and intersects it with the defect region and the bearing roller region to obtain the local background.Finally,pseudo-defect filtering is realized by screening the local background and the contrast features of the defects and the local grayscale features.The experimental results show that the algorithm can better filter the pseudo defects that appear in the detection results.After field testing,the pseudo-defect filtering based on double-threshold adaptive segmentation and optimization features can better propose complete defects and filter pseudo-defects,thus improving the extraction accuracy of defects.