Title :
Long-term coastal changes detection system based on remote sensing and image processing around an island
Author :
Bouchahma, Maged ; Yan, Wanglin
Author_Institution :
Yan Lab., SFC Keio Univ., Fujisawa, Japan
Abstract :
As an island ecosystem, Djerba, a region of Tunisia located on the southern shore of the Mediterranean Sea, is characterized by limited natural resources and threatened by land degradation due to rapid socio-economic development and heavy human-induced changes to the landscape. The objective of this study is to build a system based on computer vision and remote sensing data for monitoring changes in the coastal zones of an island. We employed monthly Landsat Thematic Mapper (TM) satellite images of the study area ranging from 1984 to 2009. The images were preprocessed using the Speeded Up Robust Features (SURF) algorithm to superimpose remote sensing images at exactly the same coordinates. We then used comparison technique to auto-validate the detection of changes. The technique is based on a window-to-window comparison of the coastal zones. Three highly affected regions were identified. The Bin El-Ouidiane (in the southeast) and Rass Errmal (in the north) regions underwent deposition during the study period, whereas the region of Rass El Kastil (in the north) underwent high erosion.
Keywords :
computer vision; erosion; feature extraction; geomorphology; geophysical image processing; geophysical techniques; oceanic crust; remote sensing; socio-economic effects; AD 1984 to 2009; Bin El-Ouidiane region; Djerba; Landsat Thematic Mapper; Mediterranean Sea; Rass El Kastil region; Rass Errmal region; Tunisia; coastal change detection system; computer vision; erosion; human-induced changes; image processing; island ecosystem; land degradation; remote sensing; socio-economic development; speeded up robust features algorithm; Canny edge detector; Coastal line change; Djerba; Landsat TM; SURF;
Conference_Titel :
Geoinformatics (GEOINFORMATICS), 2012 20th International Conference on
Conference_Location :
Hong Kong
Print_ISBN :
978-1-4673-1103-8
DOI :
10.1109/Geoinformatics.2012.6270334