The Semi-Variogram and Spectral Distortion Measures for Image Texture Retrieval

Area of Science:

• Computer Vision
• Image Processing
• Spatial Statistics

Background:

• Image texture retrieval often relies on multiple features and algorithms, posing challenges for diverse datasets.
• Single-feature classification for varied texture images remains a significant challenge in the field.

Purpose of the Study:

• To propose and evaluate a novel single-feature approach for image texture retrieval.
• To demonstrate the effectiveness of semi-variogram estimators and spectral distortion measures for texture analysis.
• To enable real-time texture retrieval using a dissimilarity measure.

Main Methods:

• Utilized semi-variogram estimators for structural and statistical texture analysis.
• Employed spectral distortion measures from linear predictive coding for signal-based similarity matching.
• Tested the approach on the Brodatz and University of Illinois at Urbana-Champaign texture databases.

Main Results:

• The proposed single-feature approach proved effective for real-time texture retrieval.
• Experimental results validated the method's performance on diverse texture databases.
• The approach offers a robust alternative to complex multi-feature retrieval systems.

Conclusions:

• Semi-variogram estimators and spectral distortion measures provide a powerful single-feature solution for image texture retrieval.
• The developed method is suitable for real-time applications requiring efficient texture analysis.
• This research advances texture retrieval by offering a theoretically sound and practically effective dissimilarity measure.