A　new　stereo　vision　calibration　algorithm　was　proposed　based　on　the　intersecting　linear　projection　and　improved　OpenCV1.0　calibration　algorithm　for　binocular　vision.　Using　OpenCV1.0　for　calibrating　often　makes　the　results　more　exaggerated.　A　new　calibrating　method　was　proposed　to　improve　the　robustness　and　real-time　of　calibration.　Precise　corner　extraction　is　very　important　as　corner　coordinate　is　known　quantity　during　algorithm　process.　Chessboard　corner　extraction　method　in　OpenCV1.0　is　used　for　corner　extraction.　Secondly,　internal　and　external　camera　parameters　of　each　camera　need　to　be　calculated　for　traditional　binocular　vision.　With　the　stereo　vision　calibration　based　on　the　intersecting　linear　projection,　internal　and　external　camera　parameters　are　not　necessary　for　calibrating.　The　proposed　method　in　this　paper　can　reduce　the　amount　of　calculation　and　running　time.　The　effectiveness　of　this　calibration　was　also　verified　by　the　experiment.