【摘要】 本文采用图像识别技术对钢铁材料在不同状态下形成的不同类型的微观组织照片进行了分析,研究了不同图像增强算法和机器学习算法对钢铁显微组织识别精度的影响。采用图像增强技术将原始图片质量提高,并采用图像特征提取技术获得图像特征。采用随机森林、支持向量机、集成树机器学习算法建立了基于图像特征的钢铁组织分类器。其中基于随机森林算法的分类器精度最高,达到了89.34%。实验表明,通过使用本研究建立的钢铁组织智能分类模型可以较为准确、高效地识别钢铁组织,可为材料设计提供参考信息。更多还原

【Abstract】 In this study,images of steel microstructures that formed in different conditions are analyzed for recognition. The influence of different image enhancement algorithms and machine learning algorithms on the accuracies of the models were studied. Image enhancement technologies were used to improve the quality of images. Features were extracted. Machine learning algorithms, including random forest,support vector machine and ensemble trees,were applied to build the classification models of microstructures based on the image features. The experimental results show that the random forest-based model has the highest accuracy 89. 34%. With the help of the model,steel microstructures could be identified accurately and efficiently,which could provide useful information for material design.更多还原