Extrinsic parameters are used in computer vision to describe the position and orientation of a camera in relation to a world coordinate system. They are crucial for tasks such as 3D reconstruction and scene understanding, as they allow for the transformation of points from the world coordinate system to the camera coordinate system.