In the rapidly expanding realm of multimedia data, information overload is an urgent issue as users increasingly seek personalized experiences. Recommendation systems aim to address this by analyzing user preferences and suggesting relevant items. However, fully capturing user interests remains a challenge, constraining system performance. Multi-behavior recommendation systems have emerged to offer more accurate recommendations by leveraging user interactions across different behaviors. This paper proposes a multi-behavior recommendation model based on hypergraph with contrastive learning. Utilizing hypergraphs, the model enriches interaction information from individual behaviors and contrasts it with graph-based information, providing the model with multi-view advantages. Through multiple layers of graph convolution, the model enriches user interests and item attributes. Finally, the feature fusion network ensures the integration of user multi-behavior information, promoting rich multi-behavior information propagation. Experiments on three real datasets demonstrate that our model outperforms leading ones, effectively overcoming limitations in multi-behavior recommendations and significantly enhancing overall recommendation performance.