Binary Pattern-Based Single-Shot Fringe Projection Profilometry for Real-Time Laparoscopic Surgical Guidance

Fringe projection profilometry (FPP) traditionally requires multi-frequency and multi-phase sinusoidal patterns and precise projector–camera synchronization, limiting real-time performance in dynamic surgical scenes. We present a compact single-shot FPP system that employs a single static binary pattern projected through the illumination channel of a dual-channel laparoscope. Depth is reconstructed from a single captured frame using a two-stage deep learning pipeline consisting of MaskNet for tissue segmentation and DepthNet for dense depth estimation. Training leverages a combination of FPP-derived ground truth and a large-scale physically based synthetic dataset generated in Blender, modeling tissue geometry, illumination, and sensor noise. Prior studies have demonstrated millimeter-level depth accuracy and improved robustness of binary patterns to motion blur and optical misalignment compared to multi-shot sinusoidal methods. In this work, we focus on the system-level design and translation of binary-pattern single-shot FPP into a compact, synchronization-free endoscopic platform, highlighting its feasibility for real-time robotic surgical guidance.