Rumor propagators are growingly employing diverse media formats to attract and deceive social media audiences. In recent years, the multimedia rumor detection model has made comprehensive use of text, visual, and social structure features for classification but does not make full use of and fusion of these features. In this paper, we put forward a Dynamic Association Hypergraph Attention Enhanced Multimodal Feature Fusion Network for rumor detection. DAHAE-MFFN enhances the representation of different patterns in a unified framework before effectively combining text, visual, and social graph features to complete the task of rumor detection. Specifically, DAHAE-MFFN introduces the Feature Information Enhancement (FIE) module, which not only enhances text and visual features by adaptively capturing the correlation between patterns but also learns complex relationships in social graphs through a dynamic association hypergraph attention network. In addition, DAHAE-MFFN introduces the Multimodal Interactive Fusion Network (MIFN) to integrate multi-modal features, which not only captures cross-modal interactions but also considers the weight allocation of different patterns at the feature level. Extensive experiments on two open datasets indicate that the algorithm surpass the existing methods in rumor detection.