Markov Random Field Segmentation for Low Frequency Active Sonar: Further Results
Abstract
The use of low frequency active sonar in shallow water leads to large numbers of clutter detections. This high false alarm rate can overload automatic tracking and classification algorithms. Traditional detection algorithms operate on each beam output individually searching for targets at all ranges. However, the target echo and bottom features may extend over several beams, either because a reflector is extended over space or because of the sidelobe structure of the beamformer. This suggests the association of detections over bearing, e.g., apply image processing to the range-bearing sonar data. A previous study described an automatic method of image segmentation based on a Markov random field (MRF) model to reduce clutter. The segmentation is treated as a labelling problem, assigning to each range-bearing cell either a target or background label, removing small objects which do not exhibit the correct signature over beams. Separate detections corresponding to one large reflector are combined and removed if they are too large to be a submarine. In this report, the algorithm is evaluated on additional data. On a single ping basis the MRF segmentation shows slightly lower detection performance at the same number of false alarm objects as the Page test detector. MRF segmentation still shows potential for fixed feature removal over multiple pings and will be studied in the feature.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2000
- Accession Number
- ADA389959
Entities
People
- R. Laterveer
Organizations
- SACLANT ASW Research Centre