Abstract: A synthetic aperture radar (SAR) target recognition method based on adaptive kernel dictionary learning is proposed in order to enhance the ability of sparse representation to extract non-linear feature information. Firstly, the SAR image feature information is mapped into a high-dimensional kernel space through a kernel function, and then the dictionary is learned in the high-dimensional kernel space. Next, the sparsity is dynamically calculated according to the information of each dictionary update. Finally, the SAR target recognition is achieved by minimizing the reconstruction error. The simulation results on MSTAR data sets show that the feature information extracted by this method can be highly indexed and has better performance on SAR target recognition.