.png){kind=icon}
.png){kind=icon}
Return
Uncertainty-Aware Vision Displacement Measurement Using a Coded Virtual Sensor Array for Outdoor Drift Mitigation
.png){kind=icon}
PRE
AI
DOI：10.1109/jsen.2026.3701434.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Long-range vision-based displacement measurement is attractive for outdoor structural monitoring, yet its accuracy is often degraded by global imaging drift induced by camera self-heating, wind-driven micromotions, and slow thermal variations. This article presents a coded virtual sensor array (CVSA)-assisted framework to suppress global drift while preserving true structural dynamics. A key practical challenge is that the near-field CVSA light-emitting diodes (LEDs) appear as defocused blobs when the camera is focused on the far-field target plane. To address this, we propose a defocus-aware parametric blob localization method that estimates subpixel LED centers and an associated covariance matrix as a confidence descriptor. The covariance information is incorporated into an uncertainty-weighted similarity alignment model to estimate the global drift transform from multiple observations, followed by temporal drift filtering to enforce slow drift evolution and mitigate intermittent outliers and overcompensation. The drift-compensated target displacement is obtained by inverse warping, and uncertainty propagation can be performed for reliability interpretation. Simulation studies validate localization accuracy and uncertainty calibration over a wide range of defocus levels. Outdoor experiments demonstrate effective reduction of low-frequency drift and long-term trends in displacement time histories while preserving dominant vibration components in the frequency domain, outperforming translation-only and unweighted similarity baselines in robustness and stability.
Keywords:
Coded virtual sensor array (CVSA)
defocused blob localization
drift suppression
outdoor structural monitoring
similarity transformation
temporal filtering
uncertainty modeling
vision-based displacement measurement
Journal
IF:
4.5
Papers:
2.1W
Citations:
7.3W
Organization
View more organizations