Abstract: A Palmprint can be divided into finger-root region, inside region and outside region with a T-structure.By appropriately utilizing the region information, the efficiency and accuracy of palmprint recognition can be improved.In this paper, an adaptive algorithm of palmprint partitioning based on T-structure is proposed.An objective function is constructed by combining the information of major lines, together with the information of grey-scale level and the orientation of the palmprint ridges.Maximizing the objective function, the T-structure can be located and the palmprint is partitioned correspondingly.Our experiment results demonstrated that the proposed algorithm is effective in producing promising partitions.