Related Experiment Video
Updated: Oct 19, 2025
12:27
Large-scale Reconstructions and Independent, Unbiased Clustering Based on Morphological Metrics to Classify Neurons in Selective Populations
Published on: February 15, 2017
7.1K
Research on a Segmentation Algorithm for the Tujia Brocade Images Based on Unsupervised Gaussian Mixture Clustering
1College of Computer Science, South-Central University for Nationalities, Wuhan, China.
Frontiers in Neurorobotics
|September 20, 2021
Summary
This study introduces an unsupervised clustering algorithm for segmenting Tujia brocade patterns, crucial for preserving cultural heritage. The method effectively extracts graphic elements by combining feature extraction, Gaussian mixture modeling, and conditional random fields for optimization.
Area of Science:
- Digital Heritage Preservation
- Computer Vision
- Cultural Informatics
Background:
- Tujia brocades are vital cultural artifacts and National Intangible Cultural Heritage, preserving the history of the Tujia nationality.
- Traditional graphic elements from Tujia brocades are essential for cultural protection and inheritance.
- Classical and deep learning segmentation methods are ineffective due to coarse textures, obvious background features, and lack of standard datasets.
Purpose of the Study:
- To develop an effective method for segmenting Tujia brocade patterns for graphic element extraction and cultural preservation.
- To address the limitations of existing segmentation algorithms for complex textile patterns.
- To create a foundation for a Tujia brocade graphic element database.
Main Methods:
- Unsupervised clustering algorithm for Tujia brocade segmentation.
- Fusion of Local Binary Patterns (LBP) and Gray-Level Co-occurrence Matrix (GLCM) to determine the optimal cluster number (K).
- Gaussian Mixture Model (GMM) for initial image clustering and segmentation.
- Voting optimization and Dense Conditional Random Field (DenseCRF) for refining segmentation results.
- Interactive cutting for final graphic element contour extraction.
Main Results:
- An effective unsupervised clustering method for segmenting Tujia brocades was developed.
- The proposed method for calculating cluster number K demonstrated ideal clustering effects.
- Voting-based optimization successfully eliminated isolated noise points in brocade patterns.
Conclusions:
- The developed unsupervised clustering approach provides a viable solution for segmenting complex Tujia brocade patterns.
- This method facilitates the extraction of graphic elements, aiding in the protection and inheritance of Tujia cultural heritage.
- The study contributes novel techniques for cluster number determination and noise reduction in image segmentation.
