structuredlight

Structured light is a 3D sensing method in which a known light pattern is projected onto a scene, typically by a projector, and captured by one or more cameras. The pattern becomes distorted by the surface geometry, and this distortion is analyzed to recover the three-dimensional shape of the object through triangulation, assuming proper calibration of the projector and cameras.

Most systems encode correspondences between projector and camera pixels using coded patterns. Gray-code sequences identify coarse correspondences, while phase-shifting fringe patterns refine depth. Some setups combine multiple patterns and use color, time multiplexing, or other encoding tricks to speed capture or improve robustness. The result is a dense 3D point cloud or mesh.

Accurate reconstruction requires calibration of intrinsics and relative pose of the projector and camera. Depth accuracy and range depend on resolution, baseline, lighting, and surface properties. Typical consumer systems operate at real-time frame rates for relatively short ranges; industrial versions can achieve higher accuracy over longer distances with specialized hardware.

Structured light is used in industrial metrology, reverse engineering, quality control, cultural heritage digitization, facial scanning, and 3D modeling of objects and scenes. It performs well on diffuse surfaces but struggles with shiny, transparent, or highly reflective materials. Ambient light and occlusions can also degrade results.

Early research dates to the 1990s. Consumer devices such as the original Microsoft Kinect and certain Intel RealSense cameras have used structured-light approaches, though newer models increasingly rely on other sensing modalities such as time-of-flight or stereo.