Multi-channel Single-Pixel Imaging for a Composite Motion Target (2504.11742v2)

Abstract: Single-pixel imaging (SPI) exhibits cost-effectiveness, broad spectrum, and stable sub-Nyquist sampling reconstruction, enabling applications across diverse imaging fields.However, due to the inherent reconstruction mechanism, SPI is not well-suited for high-speed moving targets. To address these challenges, we propose a novel, universal SPI configuration for tracking and imaging moving objects.Unlike traditional motion compensation methods, our approach enables the recovery of targets undergoing arbitrary motion, including translation, rotation, periodic, or non-periodic movements, within a two-dimensional plane without increasing the number of modulation frames.By leveraging the centroid positions from multiple wavelength channels, we determine the target's motion state from a kinematic perspective. Moreover, we developed an adapted reconstruction method, the (P-IT) pseudo-inverse transformation method, which allows for the efficient reconstruction of objects with composite motion. With a maximum flip rate of 20 kHz for the digital micromirror device (DMD), the theoretical perception frame rate can reach up to 2222 Hz, comparable to that of conventional motion-compensated SPI for purely translational objects.