Settlement extraction in the North China Plain using Landsat and Beijing-1 multispectral data with an improved watershed segmentation algorithm
- 26 March 2010
- journal article
- research article
- Published by Taylor & Francis in International Journal of Remote Sensing
- Vol. 31 (6) , 1411-1426
Abstract
In this paper we present an improved watershed segmentation algorithm for settlement mapping from medium resolution satellite data over plain areas in China. The algorithm can increase the computational efficiency of the fastest reported watershed segmentation algorithm by 30–40%. We apply this method to a selected study area in southern Hebei Province, China. We acquired a Landsat Enhanced Thematic Mapper Plus (ETM + ) image over this area in May 2000, two Landsat Thematic Mapper (TM) images in August 2004 and April 2005, and two Beijing-1 satellite images in May 2006 and May 2007. The three types of images have three similar spectral bands (green, red and near-infrared) with similar spatial resolution (30–32 m). Only the red and near-infrared bands were used in image segmentation for settlement area extraction. The extracted settlement results are compared with manual interpretation results by two people. We assumed the human interpretation results are of higher accuracy than the segmentation results. Our results indicated that our settlement area extraction method is effective. With high quality images, the overall accuracies are nearly 94%, the kappa coefficient can be greater than 0.85.Keywords
This publication has 22 references indexed in Scilit:
- An improved watershed algorithm based on efficient computation of shortest pathsPattern Recognition, 2007
- A combined fuzzy pixel-based and object-based approach for classification of high-resolution multispectral data over urban areasIEEE Transactions on Geoscience and Remote Sensing, 2003
- A multichannel watershed-based algorithm for supervised texture segmentationPattern Recognition Letters, 2003
- Use of normalized difference built-up index in automatically mapping urban areas from TM imageryInternational Journal of Remote Sensing, 2003
- An efficient watershed algorithm based on connected componentsPattern Recognition, 2000
- Integrating Contextual Information with per-Pixel Classification for Improved Land Cover ClassificationRemote Sensing of Environment, 2000
- Hybrid image segmentation using watersheds and fast region mergingIEEE Transactions on Image Processing, 1998
- A multiscale gradient algorithm for image segmentation using watersheldsPattern Recognition, 1997
- Automatic watershed segmentation of randomly textured color imagesIEEE Transactions on Image Processing, 1997
- Watersheds in digital spaces: an efficient algorithm based on immersion simulationsIEEE Transactions on Pattern Analysis and Machine Intelligence, 1991