1 January 2025 in IEEE Sensors Journal
Most infrared and visible image fusion algorithms demonstrate satisfactory performance under normal lighting conditions, but often perform poorly in low-light environments because the texture details in visible images are barely visible in the dark. To resolve this issue, we have proposed A novel fusion network based on enhancement and progressive aware for infrared – visible images in low-light (EPAFusion). The EPAFusion network efficiently brightens low-light environments while fusing complementary information from the two modalities. Specifically, we design a degraded illumination disentangled network (DIDNet), which is specially used to eliminate the degraded illumination effect in the visible image and enhance the feature information of the source image. Then, a progressive complementary aware fusion network (PCAFNet), which is used to generate fused images with rich texture details and high contrast by integrating the complementary information of the dual modal through a progressive approach. Extensive comparative experiments have demonstrated that EPAFusion exhibits superior to state-of-the-art fusion algorithms in terms of subjective and objective evaluation metrics. Particularly, low-light enhancement and dual-modal aware fusion provide more effective information to the fused image.
10 July 2025 in IEEE Sensors Journal
Fusion of complementary feature information from visible-infrared images can effectively improve the performance of target detection. However, the information of visible images is obscured in low-light environments, resulting in the unsatisfactory performance of existing visible-infrared target detection methods in low-light environments. To address this problem, we introduce dual-modality fusion detection with image-enhancement and four-head adaptively spatial feature fusion network (DFIF), which is to improve the multi-target detection accuracy in low-light environments. DFIF is able to mine more information about the target during the cross-modality fusion process through the proposed low light image enhancement network (LLIE). Furthermore, considering the differences and complementarities between the two modalities, we design a dual-modality attention feature fusion module in the dual-stream feature extraction process to fully integrate the complementary information in the feature extraction stage. Finally, we design four-head adaptively spatial feature fusion network (FASFF) with a big target detection layer that can adaptively learn spatial weights for feature mapping fusion at different scales. By optimizing the weighting parameters, FASFF effectively integrates the features at different levels to achieve higher accuracy target detection. We evaluate it on the LLVIP and M3FD datasets, and our DFIF achieves 74.1% and 73.9% on the mAP metrics, respectively, demonstrating optimal detection performance and establishing a new benchmark for subsequent dual-modality target detection.