An Improved Hierarchical Segmentation Method for Remote Sensing Images
- PDF / 900,528 Bytes
- 10 Pages / 547.087 x 737.008 pts Page_size
- 50 Downloads / 264 Views
RESEARCH ARTICLE
An Improved Hierarchical Segmentation Method for Remote Sensing Images Yumin Tan & Jianzhu Huai & Zhongshi Tang & Weiwei Xi
Received: 25 December 2009 / Accepted: 21 June 2010 / Published online: 12 February 2011 # Indian Society of Remote Sensing 2011
Abstract This paper presents an inversed quad tree merging method for hierarchical high-resolution remote sensing image segmentation, in which bottomup approaches of region based merge techniques are chained. The image segmentation process is mainly composed of three sections: grouping pixels to form image object/region primitives in imagery using inversed quad tree, initializing neighbor list and region feature variables and then hierarchical clustering neighboring regions. This segmentation algorithm has been tested on the QuickBird images and been evaluated and it exhibits good efficiency over initialization of neighbor list for quad tree node/ region primitives. This paper also provides a brief proof of the good efficiency of a sorted merge list which can be viewed as an alternative for dither matrix to randomly distribute region merging pairs which is adopted in e-Cognition. Keywords Inversed quad-tree . Image segmentation . Remote sensing . eCognition . ENVI Y. Tan (*) : J. Huai : W. Xi School of Transportation Science & Engineering, Beihang University, Xueyuan Road 37, Haidian District, Beijing 100191, People’s Republic of China e-mail: [email protected] Z. Tang Department of Civil Engineering, Tsinghua University, Beijing 100084, China
Introduction For remote sensing image analysis, image segmentation methods provide valuable alternatives to conventional per-pixel classification methods, since they take the spatial context into account (Lhermitte et al. 2008). Over the past few years, high resolution remote sensing images are sent back for processing by Quick Bird, IKONOS and other like Sensors, and many algorithms stemming from the precedent conventional ones for the purpose of segmenting high resolution images spring up like mushrooms. Generally, approaches for image segmentation fall into four categories (Cufí et al. 2003): characteristic feature threshold or clustering, edge detection, region growing or extraction, and iterative pixel classification. An image segmentation algorithm may be a combination of two or more of mentioned classes. Dividing the design of programs for image segmentation into modules may greatly improve its portability and efficiency (Zouagui et al. 2004). Combination of initial segmentation and afterwards region merging stage has obvious advantage of having initialized adjacency information before merging meaningful image object primitives produced by the first step and above all, using less process time while getting approximately the same result as expensive region merging approaches. Many preprocessing techniques have been used in color image segmentation, such as watershed transform, graph based segmentation (Bilodeau et al. 2006), naive quad
J Indian Soc Remote Sens (December 2010) 38(4):686–695
tree (Kelk
Data Loading...
