Depthmap

A depth map is a two-dimensional array in which each pixel encodes the distance from a viewpoint to the visible surface along the optical axis. Depth is typically expressed in meters, or as disparity values that can be converted to depth given camera intrinsics and baseline.

Depth maps are produced by several technologies: stereo vision uses two or more cameras to estimate disparity; structured light projects a known pattern and measures deformation; time-of-flight sensors emit light and measure return time; LiDAR systems provide high-precision depth that can be represented as depth maps; multi-view stereo reconstructs dense depth from multiple views.

Commonly stored as a grayscale image, with darker or lighter values corresponding to farther distances; zero or NaN may indicate invalid measurements. Depth values are used to reproject 2D pixels to 3D space via pinhole camera equations; the resulting point cloud can be converted to meshes, textured with color, or used directly for 3D reasoning.

Applications include robotics and autonomous navigation, augmented and virtual reality, 3D scanning, environmental mapping, gesture recognition, and computer graphics effects such as depth-based compositing or depth of field.

Limitations include noise, especially at long range or on shiny/transparent surfaces; occlusions create holes; resolution and frame rate depend on sensor; calibration is required to relate depth to real-world units. Depth maps may be incomplete or biased in certain lighting conditions.