In order to solve the problems of the feature extraction inefficiency and the noise sensitivity of traditional face recognition methods,a new face recognition algorithm based on significance diff erence local directional pattern (SDLDP) and deep convolution neural network (DCNN) is proposed in this paper.A new impro ved SDLDP face feature extraction method is given. First,the difference of eight neighboring edge respo nse values is calculated to form eight new difference directions.The corresponding directional bits of the top k difference response values are set to be 1,and those of the rest o f response values are set to be o.Then the LDP just utilizes the directions of the most prominent edge response value and t he second most prominent one.The SDLDP patterns are generated by statistical analysis acc ording to the occurrence rates of the LDP codes in a mass of face images.The face descriptor is represented by using the global concatenated histogram based on SDLDP extracted from the face image which is divided into several sub-region s.Finally,making SDLDP texture features as the input of DCNN,good network parameters are obtained by greedy training the network layer by layer.At last,the trained network is used to predict the test samples labels.The experimental results show that the algorithm can obtain higher recogn ition rate compared with traditional algorithms.