Image feature extraction and recognition model construction of coal and gangue based on image processing technology

Area of Science:

• Mining Engineering
• Computer Vision
• Image Processing

Background:

• Intelligent fully mechanized caving mining requires accurate coal gangue recognition.
• Existing methods suffer from low accuracy in identifying coal gangue.

Purpose of the Study:

• To develop an improved coal gangue recognition model for fully mechanized caving mining.
• To address the challenge of low recognition accuracy in coal gangue identification.

Main Methods:

• Designed a multi-light source image acquisition system with optimal illuminance (17,130 Lux).
• Applied Gaussian filtering to reduce noise in grayscale coal and gangue images.
• Extracted grayscale (skewness, variance) and texture (contrast) features from 900 images.
• Utilized a least squares vector machine for coal gangue identification.

Main Results:

• Identified gray skewness, gray variance, and texture contrast as highly discriminative features.
• Achieved a coal gangue recognition accuracy of 92.2% using gray skewness.
• Achieved a coal gangue recognition accuracy of 91.5% using texture contrast.

Conclusions:

• The developed model provides reliable theoretical support for accurate coal gangue recognition.
• The proposed feature extraction and recognition method significantly improves accuracy in intelligent mining.
• Optimized image acquisition and feature selection are crucial for effective coal gangue identification.