Author(s): Dongju Wu; Pieter Van Gelder; Han Vrijling
Linked Author(s):
Keywords: Dike system identification; Dike system classification; Remote sensing; Edge feature enhancement; Experience oriented classification method
Abstract: Dike systems play a significant and positive role not just in flood defense, flow diversion and irrigation, but also in traffic and landscape. The structural stability and function situation of dikes are essential for proper safety assessments. Dike systems survey in the traditional way is completed mainly through field sampling, which takes much time, money and human resources, while the results need to interpolate to cover the whole dike system and lose accuracy and local details. With the development of remote sensing (RS) techniques, high spatial resolution and high spectral resolution images are available and widely being used. Combined with the large ranges for one scope of RS image and short-cycles of repeat period for same region, it is possible to supply a more effective and convenient approach for the components identification and information extraction in a dike system. To acquire dike systems’ topographical information in large-scale, on the basic of spectral character of dike system and back ground objects, this paper first explores the filter operator of edge detection and feature extraction for sea defense in RS images to extract location line of a dike system; and then studies using land covers’ classification to get ground information nearby the dike system; and finally gives an experience oriented classification method to distinguish and locate different components of the dike system: dike, dune and hydraulic structure. The reason of choosing Landsat 7 ETM+ spectrum bands 4/3/2 and their special combination for monitoring the dike system are explained. Emissivity and spectral characteristics of different dike system components are introduced and the method chosen to distinguish different class in dike system and its limitation are discussed as well.
Year: 2013