Extracting dominant textures in real time with multi-scale hue-saturation-intensity histograms

Area of Science:

• Computer Vision
• Image Processing
• Pattern Recognition

Background:

• Extracting high-quality texture features from images is crucial but challenging due to complex color distributions.
• Existing methods for texture measurement are often laborious and computationally intensive.
• Randomness in color patterns complicates accurate texture analysis despite visual similarity.

Purpose of the Study:

• To propose an efficient and automated method for extracting dominant texture features from images.
• To overcome the difficulties associated with laborious texture measurement and complex color distribution patterns.
• To leverage the hue-saturation-intensity (HSI) color model for enhanced texture recognition.

Main Methods:

• Utilizing multi-scale color histograms to efficiently measure color distribution patterns without computing actual patterns.
• Adopting the hue-saturation-intensity (HSI) color model to incorporate human visual perception in texture recognition.
• Validating the proposed method on various benchmark datasets to assess effectiveness and efficiency.

Main Results:

• The method successfully extracts dominant texture features automatically and in real-time.
• Demonstrated significant efficiency gains, achieving speeds several orders of magnitude faster than existing approaches.
• Validated effectiveness across diverse benchmark datasets, confirming the method's robustness.

Conclusions:

• Multi-scale color histograms offer an efficient and effective solution for high-quality texture feature extraction.
• The proposed approach simplifies texture measurement, saving considerable effort and computational resources.
• This method holds promise for real-time image analysis and texture recognition applications.