Texture classification based on image (natural and horizontal) visibility graph constructing methods.

Area of Science:

• Computer Vision
• Image Analysis
• Pattern Recognition

Background:

• Texture classification is a crucial task in image analysis with broad applications.
• Existing methods often face challenges in accurately characterizing complex textures.
• The Image Visibility Graph (IVG) network construction method offers a novel approach to image representation.

Purpose of the Study:

• To introduce and evaluate a new texture classification algorithm named TCIVG (Texture Classification based on Image Visibility Graph).
• To assess the performance of TCIVG using both artificial and natural texture image datasets.
• To compare the efficacy of TCIVG against existing texture classification methods.

Main Methods:

• Texture images were transformed into natural and horizontal Image Visibility Graphs (IVGs).
• The degree distribution P(k) of these graphs was extracted as a key feature.
• Classification was performed using a quadratic discriminant for natural IVGs and a linear Support Vector Machine (SVM) for horizontal IVGs.

Main Results:

• TCIVG achieved 100% classification accuracy on artificial texture images using natural IVGs.
• TCIVG obtained 94.80% classification accuracy on natural texture images using horizontal IVGs.
• The proposed TCIVG method outperformed several existing texture classification approaches on the Brodatz database.

Conclusions:

• The TCIVG algorithm demonstrates high effectiveness for texture classification.
• Image Visibility Graphs provide a powerful feature representation for texture analysis.
• TCIVG offers a promising alternative for accurate and efficient texture classification in image analysis.