Line Stereo Matching Among Image Sequence by Using Result of Relative Orientation and Geometry Constrains from Multiply Views

This paper proposes a line stereo matching strategy, which integrates the result of relative orientation and geometry constrains. The averaged parallax which comes from homo-points in two adjacent images is used to locate the search center for target line feature. Then with the epipolar constraint existing between two views, the search range (matching window) for target line feature can be formed and the candidate lines can be selected from that special matching window.