Subpixel-Level Dynamic Displacement Tracking in Pile Driving Using Computer Vision and Frequency-Domain Matching

In response to current issues in the dynamic piledriving process, such as the reliance on manual methods for obtaining pile displacement response parameters, low accuracy, and insufficient real-time performance, this paper proposes a computer vision-based non-contact displacement monitoring and parameter extraction method. This approach utilizes the scale marks on the pile body as natural features, integrating a frequency-domain accelerated matching algorithm and quadric surface fitting technology to achieve sub-pixel-level dynamic tracking of longitudinal displacement during pile driving. Combined with a time-window step recognition algorithm, an automatic extraction mechanism for key construction parameters, such as penetration depth and rebound value, is established.