Aiming at the problem of single feature extraction by hand-designed feature methods in current face liveness detection and traditional deep learning algorithms are prone to generate local minimum and overfitting

a face liveness detection method based on local receptive field extreme learning machine（ELM-LRF）model is proposed.Firstly

the model randomly generates input weights

and then

the regularized least squares method is used to analytically calculate the output weights

which effectively solves the local minimum and over-fitting problems easily generated by deep learning algorithms and has superior generalization performance.In the CASIA-FASD and NUAA databases

the ELM-LRF model is compared with other advanced face liveness detection algorithms.The ELM-LRF algorithm has the highest classification performance.