Adaptive aperture pupil-inspired foveated single-pixel imaging for multi-posture moving targets.

The article focuses on a pupil-inspired foveated single-pixel imaging (PFSPI) technique that dynamically adjusts the shape and position of high-resolution fovea regions to improve imaging of multi-posture moving targets in dynamic scenes. Drawing inspiration from the adaptable pupil shapes found in animals, PFSPI extends traditional foveated single-pixel imaging by employing spatially variant pixel grids based on prior low-resolution measurements and object-detection algorithms to tailor illumination patterns to regions of interest (ROIs). This approach enhances effective utilization of high-resolution areas by up to 63%, improves image quality as measured by metrics like NIQE, and enables real-time tracking of target posture and movement. The method is validated in single- and multi-target scenarios, as well as complex large-scale natural scenes, with options for autonomous or manual foveal control to accommodate diverse imaging needs. Overall, PFSPI offers a flexible and practical solution for high-definition imaging and pose tracking in applications where conventional multi-pixel cameras face limitations.