Adaptive Semi-Supervised Fuzzy C-Means Method With Local Spatial Information and Pre-Clustering for Image Segmentation
The semi-supervised fuzzy C-means clustering algorithm is an improved version of the fuzzy C-means algorithm, designed to utilize a small amount of supervised information to enhance the clustering results. However, many semi-supervised fuzzy C-means algorithms suffer from the inadequate use of super...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2024-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10812752/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The semi-supervised fuzzy C-means clustering algorithm is an improved version of the fuzzy C-means algorithm, designed to utilize a small amount of supervised information to enhance the clustering results. However, many semi-supervised fuzzy C-means algorithms suffer from the inadequate use of supervised information and sensitivity to noise. Therefore, this study employs pre-clustering and label propagation to enhance efficiency of supervision and introduces spatial information to improve the robustness of algorithm to noise. First, preliminary clustering of the supervised information is conducted to distinguish feature differences within each cluster, allowing the supervised information to guide clustering more rationally. Second, supervised information is disseminated to pixels with similar features, enabling a small amount of supervised information to guide the clustering process effectively. Then, an objective function with adaptive weights is designed to calculate the weights of the local spatial information and supervision weights based on the local spatial information and label spatial information respectively, enhancing the flexibility of algorithm. Finally, experimental results on synthetic images and multiple real image datasets demonstrate that the proposed algorithm can accomplish most segmentation tasks and, in most cases, outperforms other algorithms. |
|---|---|
| ISSN: | 2169-3536 |