Type-2 fuzzy thresholding using GLSC histogram of human visual nonlinearity characteristics

Area of Science:

• Computer Vision
• Image Processing
• Fuzzy Logic

Background:

• Image thresholding is crucial for image segmentation in computer vision.
• Existing methods like traditional GLSC histograms have limitations in capturing spatial information.
• Human visual nonlinearity characteristics (HVNC) offer a new perspective for refining image analysis.

Purpose of the Study:

• To develop a new image thresholding method using type-2 fuzzy sets.
• To integrate human visual nonlinearity characteristics (HVNC) into the GLSC histogram for improved spatial information.
• To enhance image segmentation accuracy and robustness.

Main Methods:

• A novel image thresholding method based on type-2 fuzzy sets is proposed.
• The GLSC histogram is refined by embedding HVNC, considering pixel gray value and local spatial information.
• Type reduction transforms the type-2 fuzzy set to a type-1 fuzzy set for fuzziness computation.
• Optimal threshold is determined by minimizing the fuzziness of the type-1 fuzzy set.

Main Results:

• The proposed method effectively integrates pixel gray value and local spatial information.
• Experiments demonstrate the effectiveness and robustness of the new thresholding technique across various image types.
• The refined GLSC histogram with HVNC improves threshold selection accuracy.

Conclusions:

• The developed image thresholding method using type-2 fuzzy sets and HVNC-refined GLSC histograms is effective.
• This approach offers a robust solution for image segmentation in computer vision applications.
• The integration of human visual characteristics enhances the performance of image analysis techniques.