{"title":"基于状态空间模型的无人机飞机蒙皮缺陷检测系统","authors":"Mengyao Feng;Yuanming Xu;Wei Dai;Haibo Luo","doi":"10.1109/TIM.2025.3606070","DOIUrl":null,"url":null,"abstract":"This article presents an unmanned aerial vehicle (UAV)-based vision measurement system for real-time aircraft skin defect detection. Existing small-target detection algorithms often rely on transformer architectures, which, despite their accuracy, suffer from high computational complexity. To address this, we propose a novel object detection approach based on a learned neural state-space model (SSM), where image features are represented as latent dynamic systems governed by recurrent state updates rather than attention mechanisms. Experimental results show performance gains over You Only Look Once (YOLO) v8, with improvements of 2.1%, 8.1%, and 1.7% in precision, recall, and mAP50, respectively. The proposed model processes a single <inline-formula> <tex-math>$640 \\times 640$ </tex-math></inline-formula> frame in 4.2 ms (<inline-formula> <tex-math>$\\approx 238$ </tex-math></inline-formula> FPS) on an RTX 4090, demonstrating that while primarily improving detection accuracy, its linear-time complexity still ensures the real-time processing capability required for UAV-based inspection. Field tests on helicopters and transport aircraft confirm the system’s robustness, repeatability, and practical value for structural health monitoring. This work contributes a lightweight, efficient vision-based instrumentation solution incorporating dynamic modeling into the measurement process.","PeriodicalId":13341,"journal":{"name":"IEEE Transactions on Instrumentation and Measurement","volume":"74 ","pages":"1-12"},"PeriodicalIF":5.9000,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A UAV-Based Measurement System for Aircraft Skin Defect Detection Using a State-Space Model Approach\",\"authors\":\"Mengyao Feng;Yuanming Xu;Wei Dai;Haibo Luo\",\"doi\":\"10.1109/TIM.2025.3606070\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"This article presents an unmanned aerial vehicle (UAV)-based vision measurement system for real-time aircraft skin defect detection. Existing small-target detection algorithms often rely on transformer architectures, which, despite their accuracy, suffer from high computational complexity. To address this, we propose a novel object detection approach based on a learned neural state-space model (SSM), where image features are represented as latent dynamic systems governed by recurrent state updates rather than attention mechanisms. Experimental results show performance gains over You Only Look Once (YOLO) v8, with improvements of 2.1%, 8.1%, and 1.7% in precision, recall, and mAP50, respectively. The proposed model processes a single <inline-formula> <tex-math>$640 \\\\times 640$ </tex-math></inline-formula> frame in 4.2 ms (<inline-formula> <tex-math>$\\\\approx 238$ </tex-math></inline-formula> FPS) on an RTX 4090, demonstrating that while primarily improving detection accuracy, its linear-time complexity still ensures the real-time processing capability required for UAV-based inspection. Field tests on helicopters and transport aircraft confirm the system’s robustness, repeatability, and practical value for structural health monitoring. This work contributes a lightweight, efficient vision-based instrumentation solution incorporating dynamic modeling into the measurement process.\",\"PeriodicalId\":13341,\"journal\":{\"name\":\"IEEE Transactions on Instrumentation and Measurement\",\"volume\":\"74 \",\"pages\":\"1-12\"},\"PeriodicalIF\":5.9000,\"publicationDate\":\"2025-09-04\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IEEE Transactions on Instrumentation and Measurement\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://ieeexplore.ieee.org/document/11151238/\",\"RegionNum\":2,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"ENGINEERING, ELECTRICAL & ELECTRONIC\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Transactions on Instrumentation and Measurement","FirstCategoryId":"5","ListUrlMain":"https://ieeexplore.ieee.org/document/11151238/","RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}
引用次数: 0
摘要
提出了一种基于无人机的飞机蒙皮缺陷实时检测视觉测量系统。现有的小目标检测算法通常依赖于变压器结构,这种结构虽然精度高,但计算复杂度高。为了解决这个问题,我们提出了一种新的基于学习神经状态空间模型(SSM)的目标检测方法,其中图像特征被表示为由循环状态更新而不是注意力机制控制的潜在动态系统。实验结果表明,与youonly Look Once (YOLO) v8相比,性能有所提高,精度、召回率和mAP50分别提高了2.1%、8.1%和1.7%。所提出的模型在RTX 4090上以4.2 ms(约238$ FPS)的速度处理单个640$ × 640$帧,表明在主要提高检测精度的同时,其线性时间复杂性仍然确保了基于无人机的检测所需的实时处理能力。在直升机和运输机上的现场测试证实了该系统的稳健性、可重复性和结构健康监测的实用价值。这项工作提供了一种轻量级,高效的基于视觉的仪器解决方案,将动态建模纳入测量过程。
A UAV-Based Measurement System for Aircraft Skin Defect Detection Using a State-Space Model Approach
This article presents an unmanned aerial vehicle (UAV)-based vision measurement system for real-time aircraft skin defect detection. Existing small-target detection algorithms often rely on transformer architectures, which, despite their accuracy, suffer from high computational complexity. To address this, we propose a novel object detection approach based on a learned neural state-space model (SSM), where image features are represented as latent dynamic systems governed by recurrent state updates rather than attention mechanisms. Experimental results show performance gains over You Only Look Once (YOLO) v8, with improvements of 2.1%, 8.1%, and 1.7% in precision, recall, and mAP50, respectively. The proposed model processes a single $640 \times 640$ frame in 4.2 ms ($\approx 238$ FPS) on an RTX 4090, demonstrating that while primarily improving detection accuracy, its linear-time complexity still ensures the real-time processing capability required for UAV-based inspection. Field tests on helicopters and transport aircraft confirm the system’s robustness, repeatability, and practical value for structural health monitoring. This work contributes a lightweight, efficient vision-based instrumentation solution incorporating dynamic modeling into the measurement process.
期刊介绍:
Papers are sought that address innovative solutions to the development and use of electrical and electronic instruments and equipment to measure, monitor and/or record physical phenomena for the purpose of advancing measurement science, methods, functionality and applications. The scope of these papers may encompass: (1) theory, methodology, and practice of measurement; (2) design, development and evaluation of instrumentation and measurement systems and components used in generating, acquiring, conditioning and processing signals; (3) analysis, representation, display, and preservation of the information obtained from a set of measurements; and (4) scientific and technical support to establishment and maintenance of technical standards in the field of Instrumentation and Measurement.