Structural displacement estimation by fusing vision camera and accelerometer using hybrid computer vision algorithm and adaptive multi-rate Kalman filter

Abstract

Structural displacement monitoring is essential because displacement can provide critical information regarding the health condition of civil structures. However, the precise estimation of structural displacement remains a challenge. This paper describes a displacement estimation technique that fuses asynchronous acceleration and vision measurements at different sampling rates. A hybrid computer vision (CV) algorithm and an adaptive multi-rate Kalman filter are integrated to efficiently estimate high-sampling displacement from low-sampling vision measurement and high-sampling acceleration measurement. An initial calibration algorithm is proposed to automatically determine active pixels and two scale factors required in the hybrid CV algorithm without any prior knowledge or ad-hoc thresholding. The proposed technique was experimentally validated and high-sampling displacements were accurately estimated in real-time with less than 1.5 mm error, indicating the potential of the proposed technique for practical applications in long-term continuous structural displacement monitoring.