Deep photometric stereo for dynamic single-pixel 3D imaging

Real-time 3D single-pixel imaging (SPI) is challenging due to the requirement of complex calibration and the inherent tradeoff between resolution and the number of measurements. In this work, we propose a calibration-free framework that integrates binocular single-pixel imaging with a super-resolution photometric stereo network (SRPS-Net) to achieve real-time 3D SPI video. Photometric images reconstructed from arbitrary left and right viewpoints are processed by SRPS-Net to recover accurate surface normals without calibration. Experimental results show that our system achieves real-time 3D reconstruction at a resolution of 128×128 with a frame rate of 6.5 fps, reaching pixel-level accuracy. The proposed method demonstrates robust generalization to complex objects and gestures, providing a compact, cost-effective, and calibration-free solution for real-time 3D single-pixel imaging.