Self-reference location single-pixel imaging for rotating objects with multi-frame reconstruction enhancement

Rotating objects widely exist in daily scenes, yet single-pixel imaging (SPI) still faces considerable challenges in achieving high-quality images of such rotating targets. To address this issue, a self-reference location single-pixel imaging (SLSPI) method is proposed. A reference pattern is selected from the structured illumination patterns and alternately illuminates the rotation target along with the structured illumination patterns. The single-pixel detector rapidly samples the detection signals corresponding to both reference illumination patterns and structured illumination patterns, followed by Fourier filtering to suppress random noise. Benefiting from this procedure, the SLSPI method exhibits superior robustness and enables accurately locate the signals corresponding to the same position of the rotating target across different structured illumination patterns, even in cases where the target rotational speed is unknown and time-varying. This capability allows the reconstruction of multiple frames with minimal motion of the rotating target, which in turn allows for the reconstruction of high-quality images. Experimental results of imaging a rotating target verify the superiority and robustness of the proposed SLSPI method.