Size measurement of blasted rock fragments based on FRRSnet

• Published in: Measurement : journal of the International Measurement Confederation 2023-08, Vol.218, p.113207, Article 113207
• Main Authors: Guo, Qinpeng, Yang, Shijiao, Wang, Yuchen, Zhang, Huanbao
• Format: Article
• Language: English
• Subjects: Blasted rock; Distance transform; FRRSnet; Model size
• ISSN: 0263-2241
• DOI: 10.1016/j.measurement.2023.113207

•FRRSnet+, a novel network for segmenting rocks and backgrounds in image, which mainly features high segmentation accuracy and small model size.•The proposed deep learning-based blasted pile rock image segmentation method can significantly improve the segmentation accuracy.•Compared with the skip architecture, identity mapping can increase the performance of the model.•Adding ASPP based on identity mapping can significantly improve the edge identification ability of the model. To accurately segment rock particles in the blasted rock pile image, this study proposes a new segmentation method based on fully residual rock segmentation network plus (FRRSnet + ), where FRRSnet + is a new deep learning network proposed by fusing U-net, ResNet, and ASPP. The method consists of two models, FRRSnet + -1 and FRRSnet + -2, where FRRSnet + -1 is responsible for segmenting rocks and background, and FRRSnet + -2 is responsible for complementing the rock edge of FRRSnet + -1 result. Compared with U-net and Segnet showed that better results were obtained by directly using FRRSnet + -1, which had MRE, RMSE, and R2 of 2.5%, 1.55, and 99.58%, respectively. More, after using the proposed segmentation method, the segmentation results can be reduced by 1.54% and 0.77% for MRE and RMSE, respectively, improved by 0.31% for R2 compared with the results of directly using FRRSnet + -1. The proposed network and segmentation method can provide a more accurate guidance for blast parameter optimization.