Research On Metallographic Structure Analysis Technology Of W18Cr4V High Speed Steel Based On Digital Image Processing

After annealing,quenching,and tempering,W18Cr4 V high-speed steel has been widely used in industrial and agricultural fields due to its high hardness and wear resistance.However,after heat treatment,the high-speed steel is prone to uneven distribution of the metallographic structure.Therefore,detection technology must be used to analyze its uniformity and control production quality.Manual detection of high-speed steel metallographic structures has shortcomings such as low efficiency,low accuracy,and strong subjectivity.In view of this,this paper proposes to conduct research on the analysis technology of W18Cr4 V high-speed steel metallographic structure based on digital image processing.The main research work is as follows:(1)Research on image acquisition and enhancement algorithm of W18Cr4 V high-speed steel metallography.According to the inspection standards of high-speed steel,W18Cr4 V highspeed steel after heat treatment was selected to prepare metallographic samples,and UD-100 M inverted metallographic microscope was used to collect images of the metallographic structure of the sample.Then,three algorithms including histogram equalization,adaptive histogram equalization,and Contrast Limited Adaptive Histogram Equalization(CLAHE)were used to enhance the images.The results show that using CLAHE can effectively preserve the detailed features of the structure of high-speed steel metallographic organization and improve the contrast of the image,providing high-quality metallographic images for quantitative analysis of the uniformity of metallographic organization distribution.(2)Research on the denoising algorithm of W18Cr4 V metallographic images.In response to the issue of poor performance of spatial domain filtering in noise reduction for metallographic images,a wavelet denoising algorithm based on a new threshold function and an adaptive threshold is proposed.Mathematical analysis shows that the new threshold function has continuity and high-order differentiability,and its asymptotic line is closer to the original image,improving the image distortion problem and additional oscillation phenomenon caused by wavelet soft and hard threshold functions.Secondly,it is verified that the adaptive threshold is negatively correlated with the number of decomposition layers,which conforms to the law of the wavelet denoising process.After comparing the denoising results of eight algorithms,the proposed algorithm’s peak signal-to-noise ratio(PSNR)is improved by an average of 22.12%,20.36%,18.83%,12.4%,4.02%,3.65%,3.77%,and 2.69%,respectively,confirming the superiority of the proposed algorithm in denoising W18Cr4 V metallographic images.(3)Research on the segmentation algorithm of W18Cr4 V metallographic images.In response to the issues of long computation time and inaccurate results of traditional multithreshold image segmentation algorithms,a multi-threshold image segmentation method based on the Multi-Strategy Improved Sparrow Search Algorithm(MSISSA)is proposed.Three different strategies are used to improve the standard sparrow search algorithm,namely,using Tent chaotic mapping combined with reverse learning principles to enhance the quality and uniformity of the initial population,introducing adaptive weights and updating the discoverer’s position with the current global optimal position,and utilizing Cauchy mutation perturbation for local optimum solutions.Subsequently,MSISSA is combined with Otsu and Kapur entropy methods to conduct multi-threshold segmentation experiments on metallographic images.Compared with the five-threshold MSISSA-Otsu algorithm,the average processing time of the MSISSA-Kapur algorithm for segmentation images is only 0.9 seconds,and the detection efficiency is improved by an average of 16.97 times;peak signal-to-noise ratio(PSNR),structural similarity(SSIM),and feature similarity(FSIM)are improved by an average of14.04%,9.91%,and 11.10%,respectively,verifying the improved algorithm’s efficiency,accuracy,and stability in optimizing metallographic image segmentation tasks.(4)Quantitative analysis and detection system design for the uniformity of carbide metallographic structure distribution.Morphological algorithms are used to process segmented metallographic images and remove burrs and noise areas.The random block image and least squares method uniformity detection principle are applied to complete the quantitative calculation of the area ratio and distribution uniformity of the carbide metallographic structure in W18Cr4 V high-speed steel.A Matlab-based quantitative detection system for carbide uniformity is developed,which realizes functions such as denoising,segmentation,and automatic determination of structural distribution uniformity in metallographic images.Compared with manual detection,the system improves detection efficiency,accuracy,and automation level and can basically meet the requirements of carbide uniformity detection in the production process.