88.

[Cover] [Contents] [Index]

Page 177

Figure 4.16 (a) Two sizes of fuzzy partitions and their corresponding decision boundary. (b and c) Two possible shapes of decision boundary shown in bold lines.

et al., 1996; Foody, 1996) since a remotely sensed image always contains some mixed pixels, and the retrieval of the proportion of each information class within the mixed pixels can improve our image interpretation. Mixed pixel classification provides a possible solution for reducing the impact of image resolution.

The proportion of different land cover types identified at a given scale within a pixel area is a function of the spatial resolution of the sensor and the fabric of the landscape (Crapper, 1984; Campbell, 1987; see also Section 2.6.2). Since the scale of observation of land cover can range over a wide spectrum, the choice of suitable information classes for performing membership estimation is not an easy task. Another difficulty is the analysis of classification accuracy for mixed pixels because: (1) it is difficult to collect ground truth at a scale directly corresponding to remotely sensed data resolution; and (2) traditional classification accuracy analysis measurement tools may not be suitable for mixed pixel analysis. These two factors are major difficulties that limit the popularity and practical application of methods for dealing with mixed pixels. As a result, most studies for testing algorithms for mixed pixel estimation use degraded high-resolution images

[Cover] [Contents] [Index]


Classification Methods for Remotely Sensed Data
Classification Methods for Remotely Sensed Data, Second Edition
ISBN: 1420090720
EAN: 2147483647
Year: 2001
Pages: 354

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net