A robust soft margin SVM classifier based on spectral/spatial fusion: enhancement of urban classification in Riyadh, Saudi Arabia

This paper presents a robust methodology toward reliable supervised urban classification cooperatively handling spectral content from multispectral image and high resolution spatial information from QuickBird images. The main contribution is to provide a new schema for image fusion of morphological/spectral features using a soft margin SVM classifier. This investigation reproduces the spectral characteristics of original multispectral image and preserves, faithfully, the spatial information of the high resolution image. The accuracy and the effectiveness of the proposed approach were evaluated and compared with those of Maximum Likelihood (ML) and ECHO (Extraction and Classification of Homogeneous Objects) algorithms. The experiment results, conducted on urban area in Riyadh Saudi Arabia, show that the maintenance of spectral characteristics in fused images is the best, and the structure is smoothly integrated into the morphological features even for small/overlapping objects in images. This is revealed by improving the classification accuracy reaching 92%.

[1]  Jon Atli Benediktsson,et al.  Ensemble methods for spectral-spatial classification of urban hyperspectral data , 2009, 2009 IEEE International Geoscience and Remote Sensing Symposium.

[2]  Imed Riadh Farah,et al.  Multi-Spectro-Temporal Analysis of Hyperspectral Imagery Based on 3-D Spectral Modeling and Multilinear Algebra , 2013, IEEE Transactions on Geoscience and Remote Sensing.

[3]  Jon Atli Benediktsson,et al.  A new approach for the morphological segmentation of high-resolution satellite imagery , 2001, IEEE Trans. Geosci. Remote. Sens..

[4]  E. Moran.  Land Cover Classification in a Complex Urban-Rural Landscape with Quickbird Imagery. , 2010, Photogrammetric engineering and remote sensing.

[5]  A. Plaza,et al.  Semi-supervised classification of urban hyperspectral data using spectral unmixing concepts , 2013, Joint Urban Remote Sensing Event 2013.

[6]  Manfred Ehlers.  Multi-image Fusion in Remote Sensing: Spatial Enhancement vs. Spectral Characteristics Preservation , 2008, ISVC.

[7]  Paolo Gamba,et al.  Including the spatial context into decision fusion for urban area mapping using hyperspectral data , 2012, 2012 IEEE International Geoscience and Remote Sensing Symposium.

[8]  H. B. Mitchell,et al.  Multi-Sensor Data Fusion: An Introduction , 2007 .

[9]  Sayan Mukherjee,et al.  Choosing Multiple Parameters for Support Vector Machines , 2002, Machine Learning.

[10]  Liangpei Zhang,et al.  An Adaptive Multiscale Information Fusion Approach for Feature Extraction and Classification of IKONOS Multispectral Imagery Over Urban Areas , 2007, IEEE Geoscience and Remote Sensing Letters.

[11]  Xiaojun Yang,et al.  Mapping vegetation in an urban area with stratified classification and multiple endmember spectral mixture analysis , 2013 .

[12]  Sandeep Kumar,et al.  Parallel approach to expedite morphological feature extraction of remote sensing images for CBIR system , 2014, 2014 IEEE Geoscience and Remote Sensing Symposium.

[13]  F. Parmiggiani,et al.  Urban Area Classification By Multispectral SPOT Images , 1990 .