Kernel extreme learning machines for PolSAR image classification using spatial features

In this study, the impacts of polarimetrie and spatial features on the classification accuracy of full polarimetrie SAR (PolSAR) RADARSAT-2 data was investigated. Since PolSAR systems have the advantage of providing day-and-night and weather-independent images could provide the geo/bio-physical and structural information about the target objects hence are an important data source for remote sensing. PolSAR data includes geophysical(roughness and moisture), geometric(rotation, shape, size) and polarimetric as well as spatial information, as these information can be considered complementary. In this study, morphological features (opening and closing) were implemented to extract spatial features. Kernel based extreme learning machines (kELM) was used for data classification. Our results demonstrated that the classification accuracy is increased by 9.2% via inclusion of polarimetric and spatial features with highest classification accuracy was obtained as 82.61%.

[1]  Chee Kheong Siew,et al.  Extreme learning machine: Theory and applications , 2006, Neurocomputing.

[2]  Jun Li,et al.  Advanced Spectral Classifiers for Hyperspectral Images: A review , 2017, IEEE Geoscience and Remote Sensing Magazine.

[3]  Peijun Du,et al.  Random Forest and Rotation Forest for fully polarized SAR image classification using polarimetric and spatial features , 2015 .

[4]  Eric Pottier,et al.  Introduction to the Polarimetric Target Decomposition Concept , 2017 .

[5]  Heather McNairn,et al.  The Contribution of ALOS PALSAR Multipolarization and Polarimetric Data to Crop Classification , 2009, IEEE Transactions on Geoscience and Remote Sensing.

[6]  Jon Atli Benediktsson,et al.  Fusion of Support Vector Machines for Classification of Multisensor Data , 2007, IEEE Transactions on Geoscience and Remote Sensing.

[7]  Chee Kheong Siew,et al.  Universal Approximation using Incremental Constructive Feedforward Networks with Random Hidden Nodes , 2006, IEEE Transactions on Neural Networks.

[8]  Timothy A. Warner,et al.  Kernel-based extreme learning machine for remote-sensing image classification , 2013 .

[9]  Jon Atli Benediktsson,et al.  Classification and feature extraction for remote sensing images from urban areas based on morphological transformations , 2003, IEEE Trans. Geosci. Remote. Sens..

[10]  I. Hajnsek,et al.  A tutorial on synthetic aperture radar , 2013, IEEE Geoscience and Remote Sensing Magazine.

[11]  Gabriele Moser,et al.  Kernel-based classification in complex-valued feature spaces for polarimetric SAR data , 2014, 2014 IEEE Geoscience and Remote Sensing Symposium.

[12]  Olaf Hellwich,et al.  Skipping the real world: Classification of PolSAR images without explicit feature extraction , 2017 .

[13]  Yao Ding,et al.  Composite Kernel Method for PolSAR Image Classification Based on Polarimetric-Spatial Information , 2017 .

[14]  Taskin Kavzoglu,et al.  Increasing the accuracy of neural network classification using refined training data , 2009, Environ. Model. Softw..

[15]  Guang-Bin Huang,et al.  Extreme learning machine: a new learning scheme of feedforward neural networks , 2004, 2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541).

[16]  Johannes R. Sveinsson,et al.  Spectral and spatial classification of hyperspectral data using SVMs and morphological profiles , 2008, 2007 IEEE International Geoscience and Remote Sensing Symposium.

[17]  Guang-Bin Huang,et al.  Convex incremental extreme learning machine , 2007, Neurocomputing.

[18]  Guang-Bin Huang,et al.  An Insight into Extreme Learning Machines: Random Neurons, Random Features and Kernels , 2014, Cognitive Computation.

[19]  Jesús Álvarez-Mozos,et al.  On the Added Value of Quad-Pol Data in a Multi-Temporal Crop Classification Framework Based on RADARSAT-2 Imagery , 2016, Remote. Sens..