In stereo vision, disparity is the apparent shift in a scene feature’s position between images captured by two cameras from different viewpoints. A system matches the same feature in both views and measures how far apart its image positions are. With known camera geometry and calibration, that shift can be used to estimate distance: nearby objects generally show greater disparity than distant ones.
Disparity matters because depth estimates can inform robotic movement, inspection, and other automated decisions. In practice, poor lighting, weak image texture, mismatched features, or camera misalignment can make disparity unreliable and produce incorrect depth. Security and safety-minded systems should validate camera calibration and treat depth output as sensor data that may be uncertain, rather than unquestioned truth.