In recent years, researchers have focused on mitigating the impact of spectral variability (SV) on unmixing performance, leading to the development of various deep-learning-based unmixing networks. Currently, most unm...
详细信息
In recent years, researchers have focused on mitigating the impact of spectral variability (SV) on unmixing performance, leading to the development of various deep-learning-based unmixing networks. Currently, most unmixing networks that account for SV mainly rely on probabilistic generative models, which lacks specific constraints for SV and suffers from the instability of solutions generated by probabilistic models. This results in the generated endmembers or SV components lacking clear physical meaning. To avoid the problems above, we propose an endmember-oriented Transformer network (EOT-Net) that leverages the advantages of endmember bundles to introduce variability while providing stable endmember results with clear physical meaning. We design an endmember-oriented Transformer (EOT) to capture endmember-specific features through directional subspace projection and a low-redundancy attention (LRA) mechanism. Subsequently, the proposed network is divided into two branches: endmember generation and abundance estimation, to process endmember-specific features. In the endmember generation branch, endmember-specific features are transformed into intraclass weights that are used to combine signatures within the bundles, and a set of endmembers is generated for each pixel. In the abundance estimation branch, endmember-specific features are integrated using a heterogeneous information fusion (HIF) module that leverages the spatial distribution heterogeneity of the endmembers, ultimately producing the abundance results. We applied the proposed algorithm to both synthetic and real datasets, and the experimental results demonstrated the model's superiority.
Comparing images captured by disparate sensors is a common challenge in remote sensing. This requires image translation-converting imagery from one sensor domain to another while preserving the original content. Denoi...
详细信息
Comparing images captured by disparate sensors is a common challenge in remote sensing. This requires image translation-converting imagery from one sensor domain to another while preserving the original content. Denoising diffusion implicit models (DDIM) are potential state-of-the-art solutions for such domain translation due to their proven superiority in multiple image-to-image translation tasks in computer vision. However, these models struggle with reproducing radiometric features of large-scale multipatch imagery, resulting in inconsistencies across the full image. This renders downstream tasks like heterogeneous changedetection impractical. To overcome these limitations, we propose a method that leverages denoising diffusion for effective multisensor optical image translation over large areas. Our approach super-resolves large-scale low spatial resolution images into high-resolution equivalents from disparate optical sensors, ensuring uniformity across hundreds of patches. Our contributions lie in new forward and reverse diffusion processes that address the challenges of large-scale image translation. Extensive experiments using paired Sentinel-II (10 m) and Planet Dove (3 m) images demonstrate that our approach provides precise domain adaptation, preserving image content while improving radiometric accuracy and feature representation. A thorough image quality assessment and comparisons with the standard DDIM framework and five other leading methods are presented. We reach a mean learned perceptual image patch similarity of 0.1884 and a Fr & eacute;chet Inception Distance of 45.64, expressively outperforming all compared methods, including DDIM, ShuffleMixer, and SwinIR. The usefulness of our approach is further demonstrated in two Heterogeneous changedetection tasks.
Recently, Object-oriented SLAM(Object SLAM) has attracted extensive research due to its ability to perceive the environment at a 3d object level. Existing object SLAM methods mostly focus on constructing 3d object map...
详细信息
ISBN:
(数字)9798331531614
ISBN:
(纸本)9798331531621
Recently, Object-oriented SLAM(Object SLAM) has attracted extensive research due to its ability to perceive the environment at a 3d object level. Existing object SLAM methods mostly focus on constructing 3d object map for static objects or mitigating the impact of currently dynamic objects on localization and mapping. However, detection of semi-static objects whose position change while unobserved still pose a significant challenge, resulting in outdated maps, which could lead to localization and robot application failures. In this paper, we propose a method to compare current observation with the existing map, enabling the continuous detection and updating of outdated sections within the map caused by position-changing semi-static objects. First, we introduce Object Covisibility Graph(OCG) to maintain the historically observed co-visibility relationships between objects. Building on this, we design an algorithm that uses the OCG to determine whether the current camera is within the observable region of each object, and subsequently implement an object state updating algorithm to detect and update outdated sections continuously. We conduct experiments on our self-make dataset with changing objects and a dataset with only static objects. The experimental results show that our method updates the outdated parts of the map more effectively compared to previous studies.
Detecting object-level changes between two images across possibly different views (Fig. 1) is a core task in many applications that involve visual inspection or camera surveillance. Existing change-detection approache...
详细信息
ISBN:
(数字)9798331510831
ISBN:
(纸本)9798331510848
Detecting object-level changes between two images across possibly different views (Fig. 1) is a core task in many applications that involve visual inspection or camera surveillance. Existing change-detection approaches suffer from three major limitations: (1) lack of evaluation on image pairs that contain no changes, leading to unreported false positive rates; (2) lack of correspondences (i.e., localizing the regions before and after a change); and (3) poor zero-shot generalization across different domains. To address these issues, we introduce a novel method that lever-ages change correspondences (a) during training to improve changedetection accuracy, and (b) at test time, to minimize false positives. That is, we harness the supervision labels of where an object is added or removed to supervise change detectors, improving their accuracy over previous work [25] by a large margin. Our work is also the first to predict correspondences between pairs of detected changes using estimated homography and the Hungarian algorithm. Our model demonstrates superior performance over existing methods, achieving state-of-the-art results in changedetection and change correspondence accuracy across both in-distribution and zero-shot benchmarks.
Panoramic images offer a comprehensive spatial view that is crucial for indoor robotics tasks such as visual room rearrangement, where an agent must restore objects to their original positions or states. Unlike existi...
详细信息
Panoramic images offer a comprehensive spatial view that is crucial for indoor robotics tasks such as visual room rearrangement, where an agent must restore objects to their original positions or states. Unlike existing 2D scene change understanding datasets, which rely on single-view images, panoramic views capture richer spatial context, object relationships, and occlusions-making them better suited for embodied artificial intelligence (AI) applications. To address this, we introduce Panoramic Scene change Understanding (PanoSCU), a dataset specifically designed to enhance the visual object rearrangement task. Our dataset comprises 5,300 panoramas generated in an embodied simulator, encompassing 48 common indoor object classes. PanoSCU supports eight research tasks: single-view and panoramic detection, single-view and panoramic segmentation, single-view and panoramic change understanding, embodied object tracking, and change reversal. We also present PanoStitch, a training-free method for automatic panoramic data collection within embodied environments. We evaluate state-of-the-art methods on panoramic segmentation and change understanding tasks. There is a gap in existing methods, as they are not designed for panoramic inputs and struggle with varying ratios and sizes, resulting from the unique challenges of visual object rearrangement. Our findings reveal these limitations and underscore PanoSCU's potential to drive progress in developing models capable of robust panoramic reasoning and fine-grained scene change understanding.
Remote sensing has proven to be an adequate tool for observation of changes to the Earth's surface. Especially modern space-borne sensors with very-high spatial resolution offer new capabilities for monitoring of ...
详细信息
Remote sensing has proven to be an adequate tool for observation of changes to the Earth's surface. Especially modern space-borne sensors with very-high spatial resolution offer new capabilities for monitoring of dynamic urban environments. In this context, clustering is a well suited technique for unsupervised and thus highly automatic detection of changes. In this study, seven partitioning clustering algorithms from different methodological categories are evaluated regarding their suitability for unsupervised changedetection. In addition, object-based feature sets of different characteristics are included in the analysis assessing their discriminative power for classification of changed against unchanged buildings. In general, the most important property of favorable algorithms is that they do not require additional arbitrary input parameters except the number of clusters. Best results were achieved based on the clustering algorithms k-means, partitioning around medoids, genetic k-means and self-organizing map clustering with accuracies in terms of κ statistics of 0.8 to 0.9 and beyond.
This paper describes two novel learning algorithms for abrupt changedetection in multivariate sensor data streams that can be applied when no explicit models of data distributions before and after the change are avai...
详细信息
This paper describes two novel learning algorithms for abrupt changedetection in multivariate sensor data streams that can be applied when no explicit models of data distributions before and after the change are available. One of the algorithms, MB -GT, uses average Euclidean distances between pairs of data sets as the decision variable, and the other, MB - CUSUM, is a direct extension of the CUSUM algorithm to the case when the unknown probability density functions are estimated by means of kernel density estimates. The algorithms operate on a sliding memory buffer of the most recent TV data readings, and consider all possible splits of that buffer into two contiguous windows before and after the change. Despite the apparent computational complexity of O(N 4 ) of this computation, our proposed algorithmic solutions exploit the structure present in their respective decision functions and exhibit computational complexity of only O(N 2 ) and memory requirement of O(N).
The application of universal source coding algorithms to the problem of classification was initiated by Ziv and has been extended to other problems in statistics such as order estimation. In spite of the large literat...
详细信息
ISBN:
(纸本)9781424422562
The application of universal source coding algorithms to the problem of classification was initiated by Ziv and has been extended to other problems in statistics such as order estimation. In spite of the large literature, these studies have been limited to problems with fixed number of samples. In this paper we study the application of universal source coding to the problem of sequential hypothesis testing and sequential changedetection. algorithms are proposed which are inspired by Waldpsilas Sequential Probability Ratio Test (SPRT) and Pagepsilas Cumulative Sum Test (CUSUM) for these problems respectively. Performance of the proposed algorithms are studied in the asymptotic regime to demonstrate their effectiveness.
Persistent ISR (intelligence surveillance and reconnaissance) has proven its value as a tactic for national defense. This activity can collect, in particular, information necessary for executing an important concept o...
详细信息
Persistent ISR (intelligence surveillance and reconnaissance) has proven its value as a tactic for national defense. This activity can collect, in particular, information necessary for executing an important concept of operations: wide-area autonomous changedetection over long time intervals. Here we describe the remarkable potential of hyperspectral remote sensing systems for enabling such missions, using either visible or thermal infrared wavelengths. First we describe blind changedetection, in which no target knowledge is assumed. Targets that have moved can nevertheless be distinguished from naturally occurring background radiometric changes through the use of multivariate statistics informed by simple physics. detection relies on the ability of hyperspectral algorithms to predict certain conserved properties of background spectral patterns over long time intervals. We also describe a method of mitigating the most worrisome practical engineering difficulty in pixel-level changedetection, image misregistration. This has led, in turn, to a method of estimating spectral signature evolution using multiple-scene statistics. Finally, we present a signature-based detection technique that fuses two discrimination mechanisms: use of some prior knowledge of target spectra, and the fact that a change has occurred.
暂无评论