Infrared target detection algorithms play important roles in the military and civilian fields and have been widely studied. However, relatively few studies have been conducted on the use of dual-band images for targeted detection. To fully utilize the advantages of dual-band images in target detection, a fusion algorithm based on multiple features of infrared dual-band images was proposed through an in-depth analysis of the detection results. The proposed fusion algorithm utilizes a deep learning-based multi-feature fusion network to process the detection results of dual-band images, fully mine the feature information of the target, adaptively select the detection results of a single band as the output, and obtain the final decision-level fusion detection results. The experimental results show that, compared with using single-band images for object detection, the proposed infrared dual-band fusion algorithm based on multiple features can effectively utilize information from different bands, improve the detection performance, and fully leverage the advantages of infrared object detection equipment.