Image-free tracking of single-pixel detection in dynamic complex scenes

Unknown complex backgrounds degrade the tracking performance of single-pixel imaging for moving targets. The existing methods often require pre-acquisition and subtraction of the background, which is impractical in many real-world scenarios. A Fourier single-pixel imaging method that combines intermediate frequency filtering and one-dimensional optical flow is proposed to address the challenge of robustly tracking moving targets against dynamic complex backgrounds. This method initially suppresses low-frequency background interference by applying a band-pass filter directly to the sparsely sampled Fourier data, obtaining a clean projection of the target without image reconstruction. Then, a one-dimensional optical flow algorithm is subsequently designed to robustly estimate the target's displacement from these filtered projections. Our method effectively improves the tracking robustness and precision without any prior knowledge of the background while enhancing the stability of the tracking trajectory. Simulations and experiments are performed to verify the accuracy and effectiveness of the proposed method, and the results show that the proposed method improves the tracking quality while imaging moving targets against complex, low-rank backgrounds in a 512 × 512-pixel scene using a 128 × 128-pixel field of view.