A large number of facial detail features are lost in low-resolution face images, which makes the recognition rate of many classical face recognition models with good performance decrease sharply. To address this problem, this paper proposes a Hierarchical Knowledge Distillation Decoupling (HKDD) network. Firstly, the convolutional layers of the teacher network and the student network perform hierarchical feature distillation to enhance the student network's feature description ability for low-resolution samples. This distillation ensures that the low-resolution face features extracted by the intermediate layers of the student network closely approximate the high-resolution face features extracted by the intermediate layers of the teacher network, effectively transferring the powerful feature description ability of the teacher network's intermediate layers to the student network. Subsequently, the softmax layers of the teacher network and the student network perform decoupling distillation, splitting the distillation loss at the softmax layer into target class distillation loss and non-target class distillation loss. The decoupling distillation can play a guiding role in the training of student networks by leveraging the suppressed non-target distillation loss, so that the student network can learn the classification ability of universal facial features under the guidance of the teacher network, thereby ensuring that the student network has strong classification ability in unrestricted scenes. Finally, by conducting validation experiments on several low-resolution face datasets, such as TinyFace and QMUL-SurvFace, the HKDD model demonstrates superior performance in terms of recognition rate and real-time performance compared to other representative low-resolution face recognition models. The experimental results confirm the effectiveness of HKDD in handling low-resolution face recognition tasks.