基于图像分割技术的海洋用钢表征评价方法研究  

Study on Characterization and Evaluation Method of Sea Steel Based on Image Segmentation Technology

作　　者：任姿颖 赵坦 王笑辰 REN Ziying;ZHAO Tan;WANG Xiaochen(Ansteel Beijing Research Institute Co.,Ltd.,Beijing 102200,China;Ansteel Iron&Steel Research Institutes,Anshan 114009,Liaoning,China)

机构地区：[1]鞍钢集团北京研究院有限公司,北京102200 [2]鞍钢集团钢铁研究院,辽宁鞍山114009

出　　处：《鞍钢技术》2025年第2期30-36,共7页Angang Technology

摘　　要：基于对传统表征评价方法的分析,开展了基于图像分割技术的海洋用钢表征评价方法研究,提出了一种与图像分割技术相结合的海洋用钢表征评价方法。该方法采用线性迭代超像素分割算法进行预处理,并基于预处理的效果进行相的深度学习模型训练,最终形成海洋用钢表征评价方法,对海洋用钢组织进行表征评价。通过对模型的构建、训练与验证,为实现对材料性能的高效预测奠定了基础,为材料科学领域表征评价的数字化转型提供了新的思路,推动了材料的智能化进程。Based on the analysis of traditional characterization and evaluation method,the characterization evaluation method of sea steel based on image segmentation technology was studied,and a characterization evaluation method of sea steel combined with image segmentation technology was proposed.In this method,linear iterative superpixel segmentation algorithm was used for preprocessing,and deep learning model training was carried out based on the preprocessing effect,and finally a characterization evaluation method of sea steel was formed to characterize and evaluate the microstructure of sea steel.Through the construction,training and verification of the model,a foundation for the efficient prediction of material properties was layed,which provided a new idea for the digital transformation of characterization and evaluation in the field of material science,and promoted the intelligent process of materials.

关 键 词：表征评价 图像分割 材料数字化 机器学习 

分 类 号：TG1[金属学及工艺—金属学] TP3[自动化与计算机技术—计算机科学与技术]