Geodesy

Deformation monitoring

Deformation (meteorology)

Arch

Algorithm

Linguistics

Engineering

Philosophy

Arch bridge

Operating system

Bridge (graph theory)

Inclinometer

Computer science

Geology

Oceanography

Deflection (physics)

Physics

Optics

Image (mathematics)

Tracing

Feature (linguistics)

Internal medicine

Artificial intelligence

Structural engineering

Medicine

RANSAC

Computer vision

Computer-Aided Civil and Infrastructure Engineering

Completely non‐contact and fast measurements are still the focus of vision‐based monitoring in real‐bridge applications. Overall, a fast deflection measurement method, that traces the natural features of structures instead of artificial targets, was developed in this study. More specifically, due to the tiny scale change between images and the small deformation level of small and medium bridges, a simplified fast‐Hessian detector and a pre‐purification‐based RANdom Sampling Consistency (RANSAC) were proposed and verified by performing experiments under different illuminations. The non‐target deformation measurement method based on an improved algorithm was applied to field testing of an arch bridge with 100 m main girder and showed better processing speed and sub‐pixel accuracy than their originals. The optical results obtained by an industrial camera were consistent with other contact‐type sensors, such as level gauges and inclinometers.

Fast bridge deflection monitoring through an improved feature tracing algorithm