Abstract: As an emerging biometric technology, palmprint recognition has been extensively researched. Due to multiresolution characteristics of the blanket dimension (BD) and the directionality of palmprint texture, feature extraction efficiencies with blanket dimension, horizontally and vertically expanded blanket dimensions (HEBD and VEBD) are compared, and a palmprint recognition algorithm based on horizontally expanded blanket dimension is proposed according to the comparison results. The algorithm is evaluated on Hong Kong PolyU Palmprint Database V2.0. Experimental results show that the horizontally expanded blanket dimension can extract the features that well represent the palmprint, and a recognition rate of 99.9% is achieved. The total execution time of recognition is less than 287ms, which is fast enough for on-line palmprint recognition application.