Abstract: The task of fingerprint segmentation is to isolate the fingerprint area from the image background.An adaptive fingerprint segmentation algorithm based on block variance and its gradient is proposed.At first,the image is partitioned into 3×3 blocks,and then the product of the block variance and its gradient is chosen as the block feature.After that,a threshold-setting method is adopted to compute adaptively the threshold for segmentation.Finally,a morphological processing step is engaged to get the final segmented image.The experimental results on databases of FVC2002 and FVC2004 show our proposed algorithm is fast,effective and robust.