UACE2017 Proceedings: Local probability-based phase unwrapping for synthetic aperture sonar interferometry

• Session:
  Signal and image processing
• Paper:
  Local probability-based phase unwrapping for synthetic aperture sonar interferometry
• Author(s):
  Ole Jacob Lorentzen, Stig Asle Vaksvik Synnes, Roy Edgar Hansen, Torstein Olsmo Sæbø, Marc Geilhufe
• Abstract:
  Synthetic aperture sonar (SAS) increases the along-track resolution in sonar images by coherent combination of collected pings. SAS interferometry is a technique for estimating time delay differences between images at different locations and/or at different times. For a SAS system with two arrays displaced by a vertical baseline, interferometry can be used to estimate the seabed depth in high resolution. The interferogram has phase ambiguities and must be unwrapped. For wideband systems the absolute phase-difference can be estimated directly for signals without noise, but with limited signal to noise ratio we may still get phase wrap errors. Therefore, we use local phase unwrapping to correct these remaining wrap errors. We suggest an approach to the local phase unwrapping using a sliding window weighted median over the phase difference image to estimate the correct wrap for each pixel. As weights we suggest using the probability of the initially selected wrap being correct, and we also develop an estimate for this probability. We apply our method on real data from a WWII dumping site with complicated wreck structures and scattered unexploded ordinances (UXOs). The data was collected with a HISAS 1030 SAS system mounted on a HUGIN autonomous underwater vehicle. We compare the results to the initial phase-difference images and to other local phase unwrapping results.
•   Download the full paper

Contact details

• Contact person:
  Mr Ole Lorentzen
• e-mail:
  
• Affiliation:
  Norwegian Defence Research Establishment (FFI)
• Country:
  Norway