Aiming at the problems of complex finger vein images recognition such as uneven brightness

position offset

rotation

occlusion and so on

a recognition method based on improved histograms of oriented gradients（HOG） features and sparse representation is proposed.Firstly

the gradient amplitude and orientation of all pixels are calculated

and the HOG of each cell is counted; Then

the amplitude and orientation of the pixel gradient in the neighborhood of the cell are used to modify HOG of cell

and the ng HOG features fused with the neighborhood gradient information are extracted; Finally

the principal component analysis（PCA） method is used to reduce the dimension of ng HOG feature matrix

and a sparse representation super-complete dictionary is constructed. The classifier based on sparse representation is designed

and the classification results are obtained. The experimental results show that the recognition rate of the proposed method is 98. 64 % on the SDUMLA database

and the recognition time of single image with a size of 128 pixel×64 pixel is 122 ms.The recognition performance is stable in different databases such as poly U database and self-built database.