Aimed at the problem of single sample face recognition rate is not hig h,In order to improve the recognition rate of single sample face recognition,t his paper proposes a single sample face recognition method based on the combinat ion of layered LBP and HOG features of pyramid mode under submode.According to t he contribution of different parts of face to face recognition,this method firs tly extracts different parts of face through the existing classifier of differen t parts of face and divides them into different sub-images.Then,aiming at the p roblem that the number of texture features extracted by LBP descriptors is small and the image edge and orientation information cannot be well described,the me thod of fusion of layered LBP and HOG features at different levels of the pyrami d is applied to each sub-image to obtain the fusion feature vector of each sub -i mage,The Euclidean distance between the training sample and the fusion features of each sub-image corresponding to the test sample is calculated and multiplied by the pre-set weight parameters of the sub-image,then the final Euclidean di st ance is obtained by adding them together,and the id of the face is determined b y the threshold.Finally,the experiment on ORL face database shows that the prop osed method has a higher recognition rate than the existing single sample face r ecognition method,The recognition rate of ORL face database is up to 89%.