Vision-based techniques have recently gained significant attention due to their cost-effectiveness, high spatial resolution, and ability to perform full-field modal identification. However, accurately measuring sub-pixel displacements in targetless scenarios remains a considerable challenge. This paper introduces a novel population-based sub-pixel motion estimation framework that relies on block matching and does not require physical markers or predefined patterns on the monitored structures. The proposed approach employs Particle Swarm Optimization (PSO) to estimate motion in any in-plane direction, effectively overcoming difficulties related to varying textures and spatial frequencies. To enhance accuracy and address optimization-related issues, a gradient-based refinement step is integrated into the PSO algorithm. The resulting framework enables high-resolution, full-field motion estimation, which is critical for reliable modal identification. The method is validated using both synthetic and experimental datasets and shows strong performance in capturing small structural displacements, thereby advancing the capabilities of vision-based modal analysis.
