Automatic batch recognition of rock deformation areas based on image segmentation methods

Image recording and analysis is an important but time-consuming method for understanding the rock mechanics mechanism. In this paper, a method for automatic batch identification of rock deformation areas is proposed. We crop the original image to remove irrelevant background. And we use adaptive threshold segmentation, region growth segmentation and global threshold segmentation and combine the characteristics of the image to identify the rock deformation area. Finally, we use image morphology processing to make the recognition result more accurate. For validation, 359 images of the rock samples of the uniaxial compression test were quickly identified. The identification time was approximately 5 ' 56.83 '. The average relative error of the method in the X and Y directions is 10.88% and 8.60%, respectively. In addition, using the identification results and the stress-strain curve, it was found that the water content and initial crack length of rock increase, and the compressive strength decreases; the effect of the initial fracture inclination on the compressive strength of the rock is not obvious.

Automated summary

In this paper, a method for automatic batch identification of rock deformation areas is proposed. The method involves image cropping, adaptive threshold segmentation, region growth segmentation, global threshold segmentation, and image morphology processing to accurately identify the rock deformation areas. Validation using 359 rock sample images showed an average relative error of 10.88% and 8.60% in the X and Y directions, respectively. The results also revealed the impact of water content, initial crack length, and initial fracture inclination on the compressive strength of rock.