Palmprint recognition is an emerging biometric technology with many advantages. However, there is some sensitive personal information in palmprint images. Once accessed by unauthorized malicious third parties, it may lead to privacy breaches. Current palmprint recognition methods focus on recognition accuracy but ignore the privacy protection issue. In this paper, we propose a novel privacy-protecting palmprint recognition method, named Pruning Frequency Channels (PFC). Based on the difference in image perception between the human eye and the model, we eliminate the privacy information in palmprint images by pruning the low-frequency components that can be perceived by humans. Subsequently, we propose a pruning strategy for determining the pruned channels based on the energy of the frequency channels. We explore the impact of the energy ratio threshold on PFC performance and find an energy ratio threshold that is universal to different datasets. Adequate experiments conducted on multiple databases show that our PFC method can perform better than other methods and effectively hide personal privacy information with almost no damage to recognition accuracy.