| Bibliography and Index of United Nations Centre for Regional Development Publications | |||
| Citation | Kam Tin-Seong, and Ren Fuhu. “Urban Land-Use Information Extraction Using a Spatial-Based Contextual Reclassification Method.” Regional Development Studies 1 (1994/95): 213-238. | ||
|
| ||
| Year | 1995 | ||
| Author |
| ||
| Author |
| ||
| Material Type | Journal Article | ||
| Features | 50 notes; 4 tables; 5 figures (2 color images) | ||
| Pages | p. 213-238 | ||
| Relationships |
Part of
| ||
| Subjects |
CLASSIFICATION 19.01.06 COMPUTER PROGRAMS 08.15.02 INFORMATION ANALYSIS 19.01.06 LAND USE 07.05.02 MAPPING 18.07.01 REMOTE SENSING 18.04.01 | ||
| Abstract |
Past studies on the application of remote sensing for urban land-use or land-cover mapping tend to use the
terms, "land use" and "land cover," interchangeably or with no explicit distinction. The general lack of a
clear conceptual differentiation of these terms contributes to the low classification accuracy of land-use
mapping with the use of high-resolution satellite data. The term land cover relates to the type of features
present on the surface of the earth. Land use, on the other hand, is defined as man's activities which are
directly related to land. Thus, it is relatively easy to map land cover with conventional per-pixel
classification techniques because land cover is directly related to the pixel values on an image. Accurate
land-use information, however, cannot always be obtained through a direct extraction from remotely-sensed
data.
In view of this problem, a two-stage spatial-based contextual reclassification algorithm has been developed for extracting urban land-use information from high spatial resolution satellite imagery using SPOT-HRV. This article discusses the development, implementation, and evaluation of the reclassification algorithm in a microcomputer environment. The performance of this classification method is evaluated by comparing its classification accuracy with that produced by the conventional per-pixel classification method. The results reveal that the classification accuracy obtained from the spatial-based contextual classification method is significantly higher than the accuracy obtained by using the pixel-based classification method. --Journal abstract | ||
| Control No. | RDS 01k | ||