基于钢液图像的扩散效果检测方法  

Detection Method of Diffusion Effect Based on Steel Liquid Image

作　　者：鞠世龙 黎远松 Ju Shilong;Li Yuansong(School of Computer Science and Engineering,Sichuan University of Science and Engineering,Zigong 643002,China)

机构地区：[1]四川轻化工大学计算机科学与工程学院,四川自贡643002

出　　处：《洛阳师范学院学报》2024年第8期22-26,共5页Journal of Luoyang Normal University

基　　金：教育部高等教育司第二批产学合作协同育人项目(202102123021);四川省高校重点实验室基金项目(2022WYY02)。

摘　　要：针对真空浇注中钢液扩散效果检测缺乏客观性的问题,提出了颗粒度和均匀度两大检测指标,并基于YOLOv8-Seg模型建立检测流程.利用CLAHE和BPDFHE,结合一种基于低通滤波和的处理方法,对钢液图像进行预处理,并使用YOLOv8-Seg对钢液进行分割.通过计算处理后图像中的边缘像素量、二值图像像素量等参数,对钢液的颗粒度和均匀度进行检测.结果表明,通过与多名工程师的评估结果对比,该方法的结果在均匀度指标上与工程师结果较为接近,在颗粒度指标上虽然存在一定的差异但仍能接受.The researchers proposed two detection indexes of particle size and uniformity,and established a detection process based on the YOLOv8-Seg model to solve the problem of lack of objectivity in detecting the diffusion effect of steel liquid in vacuum casting.Firstly,the researchers used CLAHE and BPDFHE combined with a processing method based on low-pass filtering and sum to preprocess the steel liquid image,and used YOLOv8-Seg to segment the steel liquid.Then,the researchers detected the granularity and evenness of the liquid steel by calculating parameters such as the amount of edge pixels and the amount of binary image pixels in the processed image.The results show that,by comparing with the evaluation results of several engineers,in terms of uniformity index,the results of this method are closer to the results of engineers;in terms of the granularity index,the results are still acceptable though there are some differences.

关 键 词：真空浇注 扩散效果 限制对比度直方图均化 亮度留存动态模糊直方图均化 低通滤波 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]