DWSDiff: Dual-Window Spectral Diffusion for Hyperspectral Anomaly Detection

高光谱成像 窗口(计算) 异常检测 对偶(语法数字) 遥感 异常(物理) 扩散 计算机科学 地质学 人工智能 物理 文学类 操作系统 凝聚态物理 艺术 热力学
作者
Wenbin Chen,Xiyang Zhi,Shikai Jiang,Yuanxin Huang,Qichao Han,Wei Zhang
出处
期刊:IEEE Transactions on Geoscience and Remote Sensing [Institute of Electrical and Electronics Engineers]
卷期号:63: 1-17 被引量:8
标识
DOI:10.1109/tgrs.2025.3537331
摘要

Anomaly detection (AD) has emerged as a critical area of research in hyperspectral imagery (HIS) processing, focusing on detecting sparse, small targets with spectral and spatial features deviating from the background without prior information. The approach of AD based on reconstruction differences is a leading method in deep learning (DL) for hyperspectral AD (HAD). A key challenge is the accurate estimation of complex backgrounds. The essence of this challenge lies in accurately reconstructing background regions while inferring the latent background of anomaly regions. In this article, we propose a novel method called dual-window spectral diffusion (DWSDiff) for HAD. To address the challenge of complex background estimation in HSIs, we developed a spectral diffusion model specifically tailored for HSI. This model achieves precise background estimation through an iterative spectral diffusion and reverse reconstruction process. We also introduced a dual-window strategy to mitigate the influence of anomaly extension areas within the neighborhood on background estimation. Moreover, the scarcity of paired labeled HSIs from the same scene, with and without anomalies, limits the model’s ability to learn features between anomaly and background. To address the shortage, we devised an anomaly generation strategy based on the principal component analysis (PCA) and the linear spectral mixing model (LSMM). Building on these, we designed a training and inference framework that integrates spectral diffusion, reverse background reconstruction, and target detection. Experimental results on the Airport-Beach–Urban (ABU) hyperspectral datasets demonstrate that DWSDiff outperforms 20 state-of-the-art (SOTA) HAD methods across six different areas under the curve (AUC) metrics.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
F_ken发布了新的文献求助10
刚刚
1秒前
wyyyy发布了新的文献求助10
1秒前
if的应助被candiceli采纳,获得20
1秒前
Owen的应助被伶俐的老头采纳,获得10
1秒前
Yacoob完成签到,获得积分10
1秒前
2秒前
liu发布了新的文献求助30
2秒前
完美世界的应助被xinyang采纳,获得10
3秒前
ljf驳回了wanci的应助
3秒前
嘟嘟完成签到,获得积分10
3秒前
SciGPT的应助被阿戈美拉丁采纳,获得10
3秒前
随风完成签到,获得积分20
5秒前
Li完成签到 ,获得积分10
5秒前
大个的应助被xzl123采纳,获得10
6秒前
7秒前
7秒前
ZGD完成签到 ,获得积分10
8秒前
桃子发布了新的文献求助10
9秒前
west完成签到,获得积分10
9秒前
徐师傅完成签到,获得积分10
9秒前
小二郎的应助被312采纳,获得10
11秒前
大模型的应助被so采纳,获得10
11秒前
11秒前
章鱼小丸子完成签到 ,获得积分10
12秒前
12秒前
左旋多巴完成签到,获得积分0
12秒前
qwzh完成签到,获得积分10
12秒前
水煮电吹风的应助被hc采纳,获得10
12秒前
lydia完成签到,获得积分10
13秒前
活泼糜发布了新的文献求助10
13秒前
13秒前
13秒前
大东东完成签到,获得积分10
13秒前
14秒前
无心的小凡完成签到,获得积分10
14秒前
科研通AI2S的应助被zz采纳,获得10
14秒前
14秒前
闪闪新晴发布了新的文献求助30
14秒前
小雪要上岸完成签到,获得积分10
14秒前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
拉伸高分子熔体中缠结行为的模拟和小角中子散射研究 1700
The World Atlas of Wine 9th edition 1000
奈特基础组织学:配套组织病理学(第4版)(Netter's Essential Histology With Correlated Histopathology) 800
八旗制度の研究 600
Boundaries and Bodhisattvas 500
高分子基功能界面水蒸发材料与应用研究 500
热门求助领域 (近24小时)
化学 材料科学 医学 生物 纳米技术 计算机科学 工程类 有机化学 化学工程 生物化学 物理 复合材料 内科学 催化作用 光电子学 人工智能 细胞生物学 无机化学 物理化学 数学
热门帖子
关注 科研通微信公众号,转发送积分 8182147
求助须知:如何正确求助?哪些是违规求助? 9571790
关于积分的说明 21820037
捐赠科研通 7778601
什么是DOI,文献DOI怎么找? 3433617
关于科研通互助平台的介绍 2557786
邀请新用户注册赠送积分活动 2479652