Automatic detection and pixel-level quantification of surface microcracks in ceramics grinding: An exploration with Mask R-CNN and TransUNet

• Published in: Measurement : journal of the International Measurement Confederation 2024-01, Vol.224, p.113895, Article 113895
• Main Authors: Fu, Hui, Song, Qinghua, Gong, Jixiang, Jiang, Liping, Liu, Zhanqiang, Luan, Qiang, Wang, Hongsheng
• Format: Article
• Language: English
• Subjects: Ceramics grinding; Crack quantification; Mask R-CNN; Surface microcracks; TransUNet
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2023.113895

[Display omitted] •A pixel-level quantification model of grinding-introduced microcrack is proposed.•Using the joint Mask R-CNN and TransUNet model to detect and segment microcracks.•Challenges in quantifying size are tackled by the skeleton-based quantification model.•The evolution of microcracks with grinding parameters is quantitatively presented. Microcrack damage on the grinding surfaces of engineering ceramics is inevitably incurred. To accurately evaluate the microcrack damage, an automatic detection and pixel-level quantification model based on the joint Mask R-CNN and TransUNet is developed. In addition, a joint training strategy is employed and the model is trained effectively on the image dataset of microcrack damage derived from the Si3N4 grinding, as captured by SEM. The Mask R-CNN demonstrates reliable automatic detection of microcracks, achieving an Average Precision of AP50 = 0.989 and AP75 = 0.864. Meanwhile, the TransUNet achieves fine segmentation of microcracks with complex characteristics, with an F1 score of 0.914 and an IoU value of 0.785. A skeleton-based quantification model of microcrack size is proposed, which delivers comprehensive and precise measurements of area, length, and notably, the width. The proposed quantification model provides a technical reference for the automatic evaluation of grinding surface quality.