Real-time single-pixel imaging of moving objects using foveated modulation patterns

Single-pixel imaging (SPI) enables real-time observation of dynamic scenes by reducing the sampling rate. However, traditional methods often struggle to achieve high spatial resolution when aiming for high temporal resolution, particularly when imaging fast-moving objects. To address this challenge, we propose a real-time high-resolution imaging method for moving objects that combines Fourier modulation-based tracking with foveated modulation. In our approach, six Fourier patterns are employed to quickly locate the moving object and determine the foveal region of the field of view, after which preloaded foveated patterns are selected to perform foveated imaging. This strategy overcomes the transmission bandwidth constraint of the digital micromirror device, thereby enabling real-time monitoring. The experimental and simulation results show that our method can significantly improve the spatial resolution of the moving object compared to conventional Fourier single-pixel imaging with the same number of patterns. Furthermore, the process can also be combined with laser guidance to achieve high-resolution imaging in user-specified regions, extending the application scenarios of single-pixel imaging.