Identification And Rating Of Center Parting And Crack In The Continuous Casting Square Slab Based On Image Processing

The detection of low-fold defects in slabs is an important part of the quality inspection of slabs.At present,low-fold defect detection still relies on manual rating,which is subjective,detection efficiency and accuracy is low.In order to improve the efficiency and accuracy of low-fold defect detection of slab,this paper adopts image processing technology as the way to identify and rate defects of continuous casting billet low-fold.This topic takes center segregation and crack as the research object,the main research contents are:(1)Pretreatment of the picture of the low-strength tissue sample of the continuous casting billet.Aiming at the phenomenon of unrelated background and tangential distortion in the low-magnification picture of continuous casting billet,the pixel mapping distortion correction algorithm and the minimum rectangular segmentation method are proposed to remove the distortion of the slab sample and remove the white background.It includes Gaussian,quantization,and salt and pepper noise,and is filtered by an adaptive median reciprocal gradient weighted image filtering algorithm.(2)Identification and rating of segregation of low-fold defects in continuous casting billet.The central segregation rating parameters were quantified according to the national standard YB/T 4002-2013 "Low-casting billet low-density defect rating map".According to the position and gray scale characteristics of the segregation,the center position of the low-strength sample of the cast slab is intercepted.The gradient operator-based hybrid operator method is used to segment the gray-scale features in the image into step-type defects to remove the cracks,leaving only the suspected segregation regions.The relevant feature parameters are extracted for the suspected region,and the identified central segregation is evaluated to obtain the result.(3)Identification and rating of low-fold defect cracks in continuous casting billet.The cracking rating parameters were quantified according to the national standard YB/T 4002-2013 "Low-casting billet low-density defect rating map".According to the position.gray scale and morphological characteristics of the crack.the crack pattern containing the texture is transformed into a significant image,and the image is converted into a salient image to highlight the crack image under the texture background.For the background untextured image,the preprocessing operation is performed.The crack is then roughly separated from the slab surface using a modified Canny operator.After the interference is removed,the crack image containing the large break or the large interference point is complemented by line scan.The relevant feature parameters of the segmented suspected crack region are extracted and identified.The crack area that identifies the success is then rated according to the rating parameters.Using the above identification and rating methods,the success rate of segregation identification reached 92.5%,the success rate of rating reached 86.5%,the success rate of crack identification reached 91.2%,and the success rate of rating reached 83.9%...