Adaptive Krawtchouk-moment single-pixel imaging for a moving object

Single-pixel imaging (SPI) of dynamic scenes typically relies on reduced sampling ratios at the cost of spatial resolution. Here, we propose an adaptive SPI framework for a moving object by integrating Fourier-modulation-based localization with Krawtchouk-moment modulation. The proposed method improves the reconstruction quality of the moving object at low sampling ratios by exploiting the localized feature representation capability of discrete Krawtchouk moments and adaptively concentrating measurements within the target region. Furthermore, when the target becomes stationary, the proposed method dynamically increases the sampling ratio and progressively expands the effective imaging field of view (EFOV), enabling a transition from target-oriented local imaging to high-resolution full-field imaging. We further analyze the influence of sampling order of the Krawtchouk-moment patterns on reconstructed image quality and EFOV expansion, and formulate an optimal sampling strategy. Experimental results demonstrate that the proposed method achieves better reconstruction performance compared with conventional Fourier single-pixel imaging under low sampling conditions while maintaining full-field imaging capability. The proposed method offers an effective solution for high-fidelity single-pixel reconstruction of dynamic scenes.