A comparative study of spatial approaches for urban mapping using hyperspectral ROSIS images over Pavia City, northern Italy

Authors:
Xin Huang;Liangpei Zhang
Affiliations:
The State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, P. R. China;The State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, P. R. China
Venue:
International Journal of Remote Sensing
Year:
2009

Citing 3
Cited 2

Support-Vector Networks

Machine Learning
Mean Shift: A Robust Approach Toward Feature Space Analysis

IEEE Transactions on Pattern Analysis and Machine Intelligence
Feature Extraction Based on Decision Boundaries

IEEE Transactions on Pattern Analysis and Machine Intelligence

Object-oriented subspace analysis for airborne hyperspectral remote sensing imagery

Neurocomputing
A spatial-spectral kernel-based approach for the classification of remote-sensing images

Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

Urban mapping techniques using high spectral and spatial resolution (HSSR) data were investigated. To this aim, this paper proposes a novel mean shift (MS)-based multiscale method, and different spatial approaches are compared, including differential morphological profiles (DMPs), pixel shape index (PSI), the fractal net evolution approach (FNEA), and the proposed MS method. These spatial features were computed based on a dimensionally reduced representation that was obtained using the non-negative matrix factorization (NMF) transform. The support vector machine (SVM) was then used for classification. These algorithms were evaluated using two HSSR datasets that were obtained by using the Reflective Optics System Imaging Spectrometer (ROSIS) sensor over the urban area of Pavia, northern Italy. The results show that the spatial approaches can effectively complement the spectral features for urban mapping, and the proposed MS-based multiscale algorithm can give comparable or even better results than the FNEA, DMPs and other traditional algorithms.