Color image segmentation using adaptive hierarchical-histogram thresholding

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• Histogram-based thresholding is crucial for color image segmentation.
• Selecting optimal thresholds to differentiate objects from background is a key challenge.
• Existing methods often rely on histogram shape analysis to find thresholds at valleys.

Purpose of the Study:

• To introduce a novel hierarchical-histogram concept for multi-granularity image abstraction.
• To present the Adaptive Hierarchical-Histogram Thresholding (AHHT) algorithm for improved threshold selection.
• To evaluate AHHT's performance against established thresholding techniques.

Main Methods:

• Developed the hierarchical-histogram concept for image representation.
• Implemented the Adaptive Hierarchical-Histogram Thresholding (AHHT) algorithm.
• Compared AHHT with histon-based and roughness-index-based thresholding techniques.

Main Results:

• AHHT adaptively identifies optimal thresholds from histogram valleys.
• The algorithm achieved superior segmentation results compared to existing methods.
• AHHT demonstrated a significant reduction in time complexity.

Conclusions:

• The hierarchical-histogram approach provides an effective image abstraction.
• AHHT offers an efficient and accurate solution for color image segmentation.
• This novel method advances the field of histogram-based image analysis.