Classification of hyperspectral data over urban areas using directional morphological profiles and semi-supervised feature extraction

Wenzhi Liao, Rik Bellens, Aleksandra Pizurica, Wilfried Philips, Youguo Pi

Research output: Contribution to journalArticle

66 Citations (Scopus)

Abstract

When using morphological features for the classification of high resolution hyperspectral images from urban areas, one should consider two important issues. The first one is that classical morphological openings and closings degrade the object boundaries and deform the object shapes. Morphological openings and closings by reconstruction can avoid this problem, but this process leads to some undesirable effects. Objects expected to disappear at a certain scale remain present when using morphological openings and closings by reconstruction. The second one is that the morphological profiles (MPs) with different structuring elements and a range of increasing sizes of morphological operators produce high-dimensional data. These high-dimensional data may contain redundant information and create a new challenge for conventional classification methods, especially for the classifiers which are not robust to the Hughes phenomenon. In this paper, we first investigate morphological profiles with partial reconstruction and directional MPs for the classification of high resolution hyperspectral images from urban areas. Secondly, we develop a semi-supervised feature extraction to reduce the dimensionality of the generated morphological profiles for the classification. Experimental results on real urban hyperspectral images demonstrate the efficiency of the considered techniques.
LanguageEnglish
Pages1177-1190
Number of pages14
JournalIEEE Journal of Selected Topics in Earth Observation and Remote Sensing
Volume5
Issue number4
DOIs
Publication statusPublished - 30 Aug 2012

Fingerprint

Feature extraction
urban area
image resolution
Classifiers

Keywords

  • classification
  • high spatial resolution
  • hyperspectral data
  • morphological profiles
  • semi-supervised feature extraction

Cite this

@article{715ac7a1914f4146afa7efd397334add,
title = "Classification of hyperspectral data over urban areas using directional morphological profiles and semi-supervised feature extraction",
abstract = "When using morphological features for the classification of high resolution hyperspectral images from urban areas, one should consider two important issues. The first one is that classical morphological openings and closings degrade the object boundaries and deform the object shapes. Morphological openings and closings by reconstruction can avoid this problem, but this process leads to some undesirable effects. Objects expected to disappear at a certain scale remain present when using morphological openings and closings by reconstruction. The second one is that the morphological profiles (MPs) with different structuring elements and a range of increasing sizes of morphological operators produce high-dimensional data. These high-dimensional data may contain redundant information and create a new challenge for conventional classification methods, especially for the classifiers which are not robust to the Hughes phenomenon. In this paper, we first investigate morphological profiles with partial reconstruction and directional MPs for the classification of high resolution hyperspectral images from urban areas. Secondly, we develop a semi-supervised feature extraction to reduce the dimensionality of the generated morphological profiles for the classification. Experimental results on real urban hyperspectral images demonstrate the efficiency of the considered techniques.",
keywords = "classification, high spatial resolution, hyperspectral data, morphological profiles, semi-supervised feature extraction",
author = "Wenzhi Liao and Rik Bellens and Aleksandra Pizurica and Wilfried Philips and Youguo Pi",
year = "2012",
month = "8",
day = "30",
doi = "10.1109/JSTARS.2012.2190045",
language = "English",
volume = "5",
pages = "1177--1190",
journal = "IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing",
issn = "1939-1404",
publisher = "IEEE",
number = "4",

}

Classification of hyperspectral data over urban areas using directional morphological profiles and semi-supervised feature extraction. / Liao, Wenzhi; Bellens, Rik; Pizurica, Aleksandra; Philips, Wilfried; Pi, Youguo.

In: IEEE Journal of Selected Topics in Earth Observation and Remote Sensing, Vol. 5, No. 4, 30.08.2012, p. 1177-1190.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Classification of hyperspectral data over urban areas using directional morphological profiles and semi-supervised feature extraction

AU - Liao, Wenzhi

AU - Bellens, Rik

AU - Pizurica, Aleksandra

AU - Philips, Wilfried

AU - Pi, Youguo

PY - 2012/8/30

Y1 - 2012/8/30

N2 - When using morphological features for the classification of high resolution hyperspectral images from urban areas, one should consider two important issues. The first one is that classical morphological openings and closings degrade the object boundaries and deform the object shapes. Morphological openings and closings by reconstruction can avoid this problem, but this process leads to some undesirable effects. Objects expected to disappear at a certain scale remain present when using morphological openings and closings by reconstruction. The second one is that the morphological profiles (MPs) with different structuring elements and a range of increasing sizes of morphological operators produce high-dimensional data. These high-dimensional data may contain redundant information and create a new challenge for conventional classification methods, especially for the classifiers which are not robust to the Hughes phenomenon. In this paper, we first investigate morphological profiles with partial reconstruction and directional MPs for the classification of high resolution hyperspectral images from urban areas. Secondly, we develop a semi-supervised feature extraction to reduce the dimensionality of the generated morphological profiles for the classification. Experimental results on real urban hyperspectral images demonstrate the efficiency of the considered techniques.

AB - When using morphological features for the classification of high resolution hyperspectral images from urban areas, one should consider two important issues. The first one is that classical morphological openings and closings degrade the object boundaries and deform the object shapes. Morphological openings and closings by reconstruction can avoid this problem, but this process leads to some undesirable effects. Objects expected to disappear at a certain scale remain present when using morphological openings and closings by reconstruction. The second one is that the morphological profiles (MPs) with different structuring elements and a range of increasing sizes of morphological operators produce high-dimensional data. These high-dimensional data may contain redundant information and create a new challenge for conventional classification methods, especially for the classifiers which are not robust to the Hughes phenomenon. In this paper, we first investigate morphological profiles with partial reconstruction and directional MPs for the classification of high resolution hyperspectral images from urban areas. Secondly, we develop a semi-supervised feature extraction to reduce the dimensionality of the generated morphological profiles for the classification. Experimental results on real urban hyperspectral images demonstrate the efficiency of the considered techniques.

KW - classification

KW - high spatial resolution

KW - hyperspectral data

KW - morphological profiles

KW - semi-supervised feature extraction

UR - http://hdl.handle.net/1854/LU-2962936

U2 - 10.1109/JSTARS.2012.2190045

DO - 10.1109/JSTARS.2012.2190045

M3 - Article

VL - 5

SP - 1177

EP - 1190

JO - IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

T2 - IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

JF - IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

SN - 1939-1404

IS - 4

ER -