Search

The extended target (ET) joint detection, tracking, and classification (JDTC) algorithms based on elliptical shape utilize rough target size information for classification, which makes it difficult to effectively solve the problem of accurate classification of similarly sized targets. Therefore, this article proposes a multiple maneuvering star-convex ET JDTC algorithm called JDTC-MM-CBMeMBer filter. First, the target extent state is modeled as a star-convex shape via the star-convex random hypersurface model (RHM). By modeling the target class-related prior information with vector form, we construct its relationship with the simultaneous extent state and integrate it into the Bayesian filter framework for joint processing. Second, to solve the implementation difficulty due to the high-dimensional target state and strong nonlinear observation model in the star-convex RHM, we model the target kinematic by two separate vectors and use the particle filter (PF) to update target class probability. Next, we use the multiple model (MM) cardinality balanced multitarget multi-Bernoulli (CBMeMBer) filter to derive the JDTC recursion process of the multiple maneuvering star-convex ETs. At last, the Gamma-Gaussian –Gaussian mixture implementation is present. The simulation results show that: 1) compared with the ET JDTC algorithm based on an elliptical shape, the proposed filter can accurately classify targets with similar sizes but different shapes and 2) compared with the multiple ET tracking algorithm based on star-convex RHM, the proposed filter can significantly improve the target state estimation results without significantly increasing the running time.

The goal of hyperspectral image fusion (HIF) is to reconstruct high spatial resolution hyperspectral images (HR-HSI) via fusing low spatial resolution hyperspectral images (LR-HSI) and high spatial resolution multispectral images (HR-MSI) without loss of spatial and spectral information. Most existing HIF methods are designed based on the assumption that the observation models are known, which is unrealistic in many scenarios. To address this blind HIF problem, we propose a deep learning-based method that optimizes the observation model and fusion processes iteratively and alternatively during the reconstruction to enforce bidirectional data consistency, which leads to better spatial and spectral accuracy. However, general deep neural network inherently suffers from information loss, preventing us to achieve this bidirectional data consistency. To settle this problem, we enhance the blind HIF algorithm by making part of the deep neural network invertible via applying a slightly modified spectral normalization to the weights of the network. Furthermore, in order to reduce spatial distortion and feature redundancy, we introduce a Content-Aware ReAssembly of FEatures module and an SE-ResBlock model to our network. The former module helps to boost the fusion performance, while the latter make our model more compact. Experiments demonstrate that our model performs favorably against compared methods in terms of both nonblind HIF fusion and semiblind HIF fusion.

ABSTRACTIn recent years, deep-learning-based methods for remote sensing image interpretation have undergone rapid development, due to the increasing amount of image data and the advanced techniques of machine learning. The abundant spatial and contextual information within the images is helpful to improve the interpretation performance. However, the contextual information is ignored by most of the current deep-learning-based methods. In this letter, we explore the contextual information by taking advantage of the object-to-object relationship. Then, the feature representation of the individual objects can be enhanced. To be specific, we first build a knowledge database which reveals the relationship between different categories and generate a region-to-region graph that indicates the relationship between different regions of interest (RoIs). For each RoI, the features of its related regions are then combined with the original region features, and the fused features are finally used for object detection. The experiments conducted on a public ten-class object detection dataset demonstrate the validity of the proposed method.