The　emergent　one-dimensional　(1D)　calibration　is　very　suitable　for　multi-camera　calibration.　However　its　accuracy　is　not　satisfactory.　Conventional　optimal　algorithms,　such　as　bundle　adjustment,　do　not　perform　well　for　the　non-convex　optimization　of　1D　calibration.　In　this　paper,　a　practical　optimal　algorithm　for　camera　calibration　with　1D　objects　using　branch　and　bound　framework　is　presented.　To　obtain　the　optimal　solution　which　can　provide　epsilon-optimality,　tight　convex　relaxations　of　the　objective　functions　are　constructed　and　minimized　in　a　branch　and　bound　optimization　framework.　Experiments　prove　the　validity　of　the　proposed　method.