Local Directional Difference and Relational Descriptor for Texture Classification
The local binary pattern (LBP) has been widely used for extracting texture features. However, the LBP and most of its variants tend to focus on pixel units within small neighborhoods, neglecting differences in direction and relationships among different directions. To alleviate this issue, in this p...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-11-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/12/21/3432 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The local binary pattern (LBP) has been widely used for extracting texture features. However, the LBP and most of its variants tend to focus on pixel units within small neighborhoods, neglecting differences in direction and relationships among different directions. To alleviate this issue, in this paper, we propose a novel local directional difference and relational descriptor (LDDRD) for texture classification. Our proposed LDDRD utilizes information from multiple pixels along the radial direction. Specifically, a directional difference pattern (DDP) is first extracted by performing binary encoding on the differences between the central pixel and multiple neighboring pixels along the radial direction. Furthermore, by taking the central pixel as a reference, we extract the directional relation pattern (DRP) by comparing binary encodings representing different directions. Finally, we fuse the above DDP and DRP to form the LDDRD feature vector. Experimental results on six texture datasets reveal that our proposed LDDRD is effective and outperforms eight representative methods. |
|---|---|
| ISSN: | 2227-7390 |