Application of improved and efficient image repair algorithm in rock damage experimental research

Area of Science:

• Geotechnical Engineering
• Materials Science
• Artificial Intelligence

Background:

• Digital image and acoustic emission technologies have limitations in rock damage analysis due to data loss and signal interference.
• Current monitoring errors can compromise the accuracy of rock damage assessments in industrial applications.

Purpose of the Study:

• To develop a robust method for restoring distorted or missing image data in rock mechanics studies.
• To improve the accuracy and efficiency of rock damage analysis using enhanced digital image processing.

Main Methods:

• An improved Incremental Transformer algorithm was utilized for repairing strain nephograms from digital image technology.
• Deep separable convolutional networks were implemented to optimize computational efficiency.
• A comprehensive training dataset of strain nephograms was generated for model training and validation.

Main Results:

• The developed algorithm successfully restored full strain detail in damaged rock images.
• Analysis using repaired images showed a closer correlation to actual rock damage processes.
• The enhanced method demonstrated superior performance compared to conventional digital image and acoustic emission techniques.

Conclusions:

• The improved Incremental Transformer algorithm offers an innovative approach to traditional rock damage analysis.
• This technique enhances the efficiency and reliability of digital image technology in rock mechanics.
• The findings contribute to cost and time savings in the petroleum and coal industries.