Detailed mapping of shallow-water environments using image texture analysis on sidescan sonar and multibeam backscatter imagery

VA Huvenne, V Huenerbach, P Blondel, TP LeBas

Research output: Contribution to conferencePaper

Abstract

Multibeam backscatter (240 kHz) and sidescan sonar data (325 kHz) of the West Solent (Southern UK) were analysed with automated image classification techniques based on Grey Level Co-occurrence Matrices (GLCM?s) to assist the mapping of different seabed facies in this dynamic, tide-influenced area. After processing and ortho-rectification with PRISM, the GLCM entropy and homogeneity indices were calculated with TexAn. Those image texture indices quantify the lack of organisation and the amount of local similarities in a chosen neighbourhood around each pixel. Used together with the original images, they help the human eye to distinguish different patterns and hence to increase the understanding of the morphology and dynamic processes at the seabed. The fact that the multibeam and sidescan sonar data were collected simultaneously allowed a comparison between the two survey methodologies and allowed to test the performance of the GLCM analysis for the two types of imagery. The sidescan sonar gives a much wider range and hence a much better coverage than the multibeam, which is also affected by high levels of speckle noise and lesser quality imagery in the far ranges. In addition, due to the different geometry and beam-forming patterns of the 2 methods, similar features may yield different textural complexities, which makes comparison between the two classification results not so straightforward.
Original languageEnglish
Publication statusPublished - Jun 2007
EventUnderwater Acoustic Measurements 2007 - Heraklion, Crete
Duration: 1 Jun 2007 → …

Conference

ConferenceUnderwater Acoustic Measurements 2007
CityHeraklion, Crete
Period1/06/07 → …

Fingerprint

Dive into the research topics of 'Detailed mapping of shallow-water environments using image texture analysis on sidescan sonar and multibeam backscatter imagery'. Together they form a unique fingerprint.

Cite this