UACE2017 Proceedings: Sound speed correction using autofocus on SAS images



    • Session:
      Towards Automatic Target Recognition. Detection, Classification and Modelling
    • Paper:
      Sound speed correction using autofocus on SAS images
    • Author(s):
      Ulrich Herter, Holger Schmaljohann, Thomas Fickenscher
    • Abstract:
      The quality of automatic target recognition (ATR) results depends heavily on the quality of the images it works on.\nOne source of quality loss is using an inaccurate sound speed during synthesizing the aperture.\nSAS images usually comprise a long abeam range whereas the sound speed is measured on the sonar platform directly and/or using CTD profiles taken at few, selected locations.\nThis makes it difficult to get valid sound speed values for the entire area imaged.\nErrors in sound speed do not only cause wrong range estimates but also blurring in along track direction due to an erroneous range migration correction.\nDeviations as little as 1% from the true sound speed may lead to significant image degradation.\nThe blurring forms a characteristic hyperbolic shape which can be exploited to derive an improved average sound speed value.\n\nWe present a method for sound speed correction employing a modified stripmap phase gradient autofocus (SPGA) and a previously published method employing phase gradient autofocus (PGA) and compare the two variants:\nWe used SPGA or PGA to estimate phase error histories from the image data itself and fited error models to the estimates.\nThe fit parameters allowed us to calculate an improved sound speed.\nAfter applying an inverse wavenumber algorithm with the old, erroneous sound speed, the new, corrected sound speed value was applied to the artificial raw data by forward wavenumber processing.\n\nA comparison of SPGA and PGA based sound speed correction is shown on simulated as well as on field data comprising different features (a wreck, ripples and stones):\nThe original images were processed using deliberately biased sound speeds as well as the best available values from the missions.\nIn most cases, our correction method was able to significantly improve the images.\nEven in cases where the best available sound speed was used in the first place, we achieved improvements.
    •   Download the full paper

    Contact details

    • Contact person:
      Mr Ulrich Herter
    • e-mail:
    • Affiliation:
      Helmut-Schmidt-Univerity Hamburg
    • Country:
      Germany