Research on target detection and segmentation in forward looking multi-beam sonar images
Zhang, D.; Fan, W.; Zeng, S. (2019). Research on target detection and segmentation in forward looking multi-beam sonar images, in: 2019 IEEE International Conference on Signal, Information and Data Processing (ICSIDP), Chongqing, China, December 11-13, 2019. pp. 1-5. https://dx.doi.org/10.1109/icsidp47821.2019.9173403
|
| Available in | Authors |
|
Document type: Conference paper
|
| Author keywords |
Multi-beam, sonar images, detection and segmentation, one-dimensional search, depth-first search |
| Authors | | Top |
- Zhang, D.
- Fan, W.
- Zeng, S.
|
|
|
| Abstract |
Due to the effect of complex environment and other interference, sonar images are always filled up with contamination and noise. For that reason, accurate and fast detection and segmentation of targets in sonar images are vital for underwater target recognition. In this paper an algorithm based on depth-first search is proposed for target detection and segmentation in forward looking multi-beam sonar images. The algorithm is made up with one-dimensional peak search and two-dimensional depth-first search segmentation. It can rapidly detect and segment the parts of targets in sonar images. The algorithm is tested with data from an actual multi-beam sonar. It is also compared with other segmentation method such as Otsu, C-means and Markov Random Fields. The result shows that the algorithm can extract the target with high efficiency while automatically avoiding reverberation and noise, making it suitable for real-time process of sonar images with isolated targets. |
|
