Fusion of Polarimetric Features and Structural Gradient Tensors for VHR PolSAR Image Classification

Abstract

This article proposes a fast texture-based supervised classification framework for fully polarimetric synthetic aperture radar (PolSAR) images with very high spatial resolution (VHR). With the development of recent polarimetric radar remote sensing technologies, the acquired images contain not only rich polarimetric characteristics but also high spatial content. Thus, the notion of geometrical structures and heterogeneous textures within VHR PolSAR data becomes more and more significant. Moreover, when the spatial resolution is increased, we need to deal with large-size image data. In this work, our motivation is to characterize textures by incorporating (fusing) both polarimetric and structural features, and then use them for classification purpose. First, polarimetric features from the weighted coherency matrix and local geometric information based on the Di Zenzo structural tensors are extracted and fused using the covariance approach. Then, supervised classification task is performed by using Riemannian distance measure relevant for covariancebased descriptors. In order to accelerate the computational time, we propose to perform texture description and classification only on characteristic points, not all pixels from the image. Experiments conducted on the VHR F-SAR data as well as the AIRSAR Flevoland image using the proposed framework provide very promising and competitive results in terms of terrain classification and discrimination.

A Compound Polarimetric-Textural Approach for Unsupervised Change Detection in Multi-Temporal Full-Pol SAR Imagery

“…1 shows the block scheme for the proposed approach. Four main stages are composing the scheme, namely: i) generation of the multi-look Coherency matrix for single-time images; ii) generation of the texture descriptors for the single-time images, based on the De Zenzo gradient tensor [10]; iii) definition of the CI feature; iv) thresholding and generation of the CD map.…”

“…However, texture represents an additional information source for target discrimination in SAR imagery. Some of state-of-the-art texturebased classification approaches considers the combination of the radiometric information with textures from Gray Level Co-occurrence Matrix (GLCM) [7], Gabor filters [8] or gradient tensors [9,10]. To the best of our knowledge, a very small effort has been spent in combining radiometric and textural information for CD in PolSAR data.…”

“…Next, a feature vector for the two images is built by considering both the radiometric and the textural information The radiometric information is given by the multi-looked terms of the covariance matrix. The textural information is given by the gradient tensors of the covariance matrix, extracted based on the min-ratio operator along multiple directions [10]. A statistical distance between the two feature vectors is defined as a Change Index (CI) and thresholded.…”

A SAR-GMTI Approach Aided by Online Knowledge With an Airborne Multichannel Quad-Pol Radar System

“…2) can be regarded as the spatial structure features, such as color features [46] and texture features [47]. In this paper, the image gradients are utilized to represent the texture features with ,,, are calculated using mean ratio operator [48].…”

Joint Polarimetric-Adjacent Features Based on LCSR for PolSAR Image Classification

“…PTD as a widely used feature extraction method, is able to decompose PolSAR images into different scattering components that can intrinsically characterize the scattering properties of different grounds. PolSAR image classification can be achieved by directly or indirectly combining these different scattering features with classifiers, such as [4]- [6]. The methods based on machine learning introduce some mature classification methods in other fields for PolSAR image, and improve the methods to be applicable to PolSAR image, such as [7] [8].…”