Self-attention-based approaches that leverage global context information for hyperspectral image (HSI) classification have gained increasing prominence. Nevertheless, due to the assignment of equivalent attention weight to all the tokens (pixels or patches), the existing self-attention mechanism inadvertently prioritizes the non-label-specified information over the instinct label-specified information, which generates attention shifts and redundancy in HSI classification. To alleviate the mentioned barrier, we propose the center-specific perception transformer network (CP-Transformer), which is the first attempt to perform class-guided attention and filter interference factors for HSI classification feature representation. Specifically, the central-pixel focus attention module (CFA) is presented to compute the label-related attention between the center and other pixels. In this manner, CFA reduces computational complexity and closely aligns with the center-pixel labeling strategy. Besides, the spectral saliency focus attention module (SSFA) is developed to capture the spectral correlation by focusing salient bands to provide a beneficial supplement for spatial features. Moreover, the hierarchical integration network (HIN) constructs the inference network to integrate and rectify spatial-spectral features for HSI classification. The experiment results on four popular HSI datasets demonstrate that the proposed method achieves robust performance compared to other state-of-the-art methods. Our code will be released at https://github.com/Chirsycy/CP-Transformer.

Recently, the methods based on self-attention mechanisms have gained increasing prominence in hyperspectral image classification (HSIC). However, the existing self-attention mechanism suffers the challenge of attention shift and redundancy. To address the problem, we propose the center category focusing transformer network (CCSF-Transformer) for HSIC, which is designed to resolve attention shifts and redundancy by balancing the multiple category features. Specifically, the central-category-focused attention mechanism (CFA) is presented in the proposed framework to compute the category-matched attention between the center pixel and neighbor pixels, closely matching the center-pixel style labeling strategy, and reducing the computation complexity by excluding the computation between interference pixels. Besides, the spectral-salient-focused attention module (SFA) is developed to capture the spectral correlation, which concentrates on the salient bands and suppresses the expression of redundant bands. Moreover, the hierarchical integration network (HIN) is built to rectify spatial and spectral features The experiment results on two popular HSI datasets demonstrate that the proposed method achieves robust performance compared to other state-of-the-art methods.

Hyperspectral image classification (HSIC) models have made remarkable progress in the last decade. Nevertheless, the downsized mapping in the convolutional neural network (CNN) and down-sampled mechanism in the transformer-based approach amplify the loss of hidden knowledge in the subpixel that encompasses crucial yet unseen features within a single pixel. Considering this aspect, the mentioned popular solutions for HSIC contradict the inherent characteristic of hyperspectral data. To address this issue, we rethink the size factor in CNN and propose a novel spatial mapping expansion with spectral compression (SMESC) network for HSIC. Specifically, the SMESC builds a mapping expansion network to mine unseen information in subpixels with enlarged feature maps. A channel modulation residual block (CMRB) is developed to compress spectral redundancy and promote salient channels with modulation information. Moreover, we design a multiple-size training strategy to substitute the traditional multiple feature extraction (FE) branches and improve the model adaptation to the different sizes of the testing samples. The extensive experimental results and analysis of four hyperspectral image (HSI) datasets demonstrate the superiority of the proposed architecture compared to other advanced HSIC methods. Our code will be released athttps://github.com/Chirsycy/SMESC.

Sensor network localization is an instance of the NP-Hard graph realization problem. Thus, methods used in practice are not guaranteed to find the correct localization, even if it is uniquely determined by the input distances. In this article, we show the following: if the sensors are allowed to wiggle, giving us perturbed distance data, we can apply a novel algorithm to realize arbitrary Generically Globally Rigid graphs (GGR), or certain vertex subsets in non-GGR graphs whose relative positions are fixed (which include vertex sets of GGR subgraphs). And this strategy works in any dimension. In the language of structural rigidity theory, our approach corresponds to calculating the approximate kernel of a generic stress matrix for the given graph and distance data. To make our algorithm suitable for real-world applications, we also present: (i) various techniques for improving the robustness of the algorithm in the presence of measurement noise; (ii) an algorithm for detecting certain subsets of graph vertices whose relative positions are fixed in any generic realization of the graph and robustly localizing these subsets of vertices, (iii) a strategy for reducing the number of measurements needed by the algorithm. We provide simulation results of our algorithm.

Sensor network localization is an instance of the NP-HARD graph realization problem. Thus, methods used in practice are not guaranteed to find the correct localization, even if it is uniquely determined by the input distances. In this paper, we show the following: if the sensors are allowed to wiggle, giving us perturbed distance data, we can apply a novel algorithm to realize arbitrary generically globally rigid (GGR) graphs (or maximal vertex subsets in non-GGR graphs whose relative positions are fixed). And this algorithm works in any dimension. In the language of structural rigidity theory, our approach corresponds to calculating the approximate kernel of a generic stress matrix for the given graph and distance data. To make our algorithm suitable for real-world application, we present techniques for improving the robustness of the algorithm under noisy measurements, and a strategy for reducing the required number of measurements.

We present a new algorithm for view-dependent level-of-detail rendering of meshes. Not only can it effectively resolve complex geometry features similar to edge collapse-based schemes, but it also produces meshes that modern graphics hardware can render efficiently. This is accomplished through a novel hybrid approach: For each frame, we view-dependently refine the progressive mesh (PM) representation of the original mesh and use the output as the base domain of uniform regular refinements. The algorithm exploits frame-to-frame coherence and only updates portions of the output mesh corresponding to modified domain triangles. The PM representation is built using a custom volume preservation-based error function. A simple k-d tree enhanced jump-and-walk scheme is used to quickly map from the dynamic base domain to the original mesh during regular refinements. In practice, the PM refinement provides a view-optimized base domain for later regular refinements. The regular refinements ensure almost-everywhere regularity of output meshes, allowing optimization for vertex cache coherence and caching of geometry data in high-performance graphics memory. Combined, they also have the effect of allowing our algorithm to operate on uniform clusters of triangles instead of individual ones, reducing CPU workload.