In recent years, the number of ransomware- related cyberattacks has dramatically increased, posing an escalating threat to cybersecurity. Therefore, this paper proposes a multi-view feature fusion approach aimed at improving the accuracy and robustness of ransomware detection. Firstly, distinct features from executable files, including dynamic, static, and image features, are extracted separately. These features are then fused to construct a comprehensive feature vector. Simultaneously, a weight self- learning mechanism is introduced during the feature fusion process to dynamically adjust the weights of different features, reducing noise and interference from irrelevant features. Finally, through an evaluation on a real ransomware dataset, we demonstrate the effectiveness of the proposed method. Experimental results indicate a significant performance improvement in ransomware detection tasks using this approach.