Adaptive Fuzzy Learning Superpixel Representation for PolSAR Image Classification

Abstract

The increasing applications of Polarimetric SAR (PolSAR) image classification demand for effective superpixels algorithms. Fuzzy superpixels algorithms reduce the misclassification rate by dividing pixels into superpixels, which are groups of pixels of homogenous appearance, and undetermined pixels. However, two key issues remain to be addressed in designing a fuzzy superpixel algorithm for PolSAR image classification. Firstly, the polarimetric scattering information, which is unique in PolSAR images, is not effectively used. Such information can be utilized to generate superpixels more suitable for PolSAR images. Secondly, the ratio of undetermined pixels is fixed for each image in the existing techniques, ignoring the fact that the difficulty of classifying different objects varies in an image. To address these two issues, we propose a polarimetric scattering information based adaptive fuzzy superpixels (AFS) algorithm for PolSAR images classification. In AFS, the correlation between pixels' polarimetric scattering information, for the first time, is considered to generate superpixels. This correlation is further used to dynamically and adaptively update the ratio of undetermined pixels. AFS is evaluated extensively against different evaluation metrics and compared with the state-of-the-art superpixels algorithms on three PolSAR images. The experimental results demonstrate the superiority of AFS on PolSAR image classification problems.

Multiobjective Evolutionary Superpixel Segmentation for PolSAR Image Classification

“…Li et al [29] proposed a new cross-iterative strategy for PolSAR superpixel segmentation that combined an improved Wishart distance and geodesic distance to generate stable superpixels with a high boundary recall rate (BR). Guo et al [30] proposed an adaptive fuzzy super-pixel segmentation method that introduced the correlation of polarization scattering information into pixels and adjusted the proportion of undetermined pixels adaptively. Although these superpixel segmentation techniques can achieve good performance for PolSAR images, most of them need to determine the number of superpixels manually in advance.…”

Design of a Hybrid GWO CNN Model for Identification of Synthetic Images via Transfer Learning Process

“…These models showcase low complexity, but cannot be scaled for multiple numbers of classes. To enhance this performance, work in [11,12,13,14] proposes use of adaptive fuzzy learning (AFL), active ensemble deep learning (AEDL), autoencoder regularization joint contextual attention network (ARJCAN), which assists in improving classification performance for multiple datasets and scenarios. Similar models are discussed in [ 15,16,17], which propose use of Spatial & Semantic Features, Novel Attention Fully Convolutional Network Method (NAFCNN), which allow the model to augment multiple feature sets for enhancing classification performance.…”

“…Most current techniques are tailored to specific types of synthetic images, such as those generated by Generative Adversarial Networks (GANs) [26]. This limits their ability to identify synthetic images generated by other types of models or techniques [11,12].…”

Algoritmo de Filtro de Kalman Baseado em Modelo Fuzzy Tipo-2 Evolutivo para Rastreamento

“…A modelagem fuzzy a partir de dados experimentais tem sido largamente desenvolvida e aplicada com sucesso em diferentes domínios de aplicac ¸ão, tais como biomédica [1], epidemiologia [2], processamento digital de imagem [3], reconhecimento de padrão [4], controle e automac ¸ão de sistemas dinâmicos [5], [6], entre outros. O interesse crescente da comunidade científica pelos sistemas fuzzy é devido a algumas de suas vantagens sobre outras técnicas de inteligência artificial, tais como a sua capacidade de combinar o conhecimento do especialista com informac ¸ões extraídas de dados experimentais, sua fácil formulac ¸ão e implementac ¸ão, maior interpretabilidade e alta precisão no tratamento de não-linearidades e incertezas [7].…”