StereoKameras

StereoKameras are imaging systems that use two or more horizontally aligned cameras to capture scenes from slightly different viewpoints, enabling depth perception and 3D reconstruction. Typically the cameras are mounted with a fixed baseline and synchronized exposure, producing simultaneous left and right images. Depth is recovered by computing the disparity between corresponding points in the two images; closer objects have larger disparities.

Key processing steps include stereo rectification to align image rows, stereo matching or correspondence search, and triangulation to convert disparity to depth. Along with passive stereo (no active illumination) there are active stereo approaches that project a known pattern to improve matching in textureless regions.

Common outputs are disparity maps, depth maps, and 3D point clouds; some systems provide synchronized color data, yielding RGB-D data.

Configurations vary: parallel, where optical axes are nearly parallel, or convergent, where cameras are angled toward a convergence point. Calibration yields intrinsic parameters (focal length, principal point, distortion) and extrinsic parameters (rotation and translation between cameras), notably the baseline distance.

Applications include robotics and autonomous navigation, 3D mapping and surveying, industrial inspection, and augmented or mixed reality where real-time depth information enhances interaction and occlusion handling.

Advantages include relatively low hardware cost and passive sensing; limitations include sensitivity to lighting, occlusions, textureless regions, reflective surfaces, and high computational demands for dense depth maps.

History: stereo vision concepts date to the 19th century (binocular disparity), with significant advances in the late 20th century onward enabling practical real-time stereo systems using digital cameras and commercial algorithms.