FusionIPCS: Infrared and visible light image fusion through an intelligent parallel control system

•Explored the fusion mechanism and drew conclusions.•Constructed a new fusion framework to allow users to adjust the results.•Utilized predecessors' experience to construct an expert system and improve the robustness of the fusion system. Different imaging mechanisms exist for infrared and visible light images with a strong complementarity between information. However, existing infrared and visible light image fusion methods rarely consider the optimal control effect of the fusion model. Accordingly, the fusion model cannot adaptively adjust and generate the optimal fusion result according to user requirements. In response to this issue, this article proposes a fusion method for infrared and visible light images based on a parallel fuzzy PID controller called FusionIPCS. First, the multi-scale fusion model is employed to optimize the fusion model from feature extraction and the construction method of feature layer and weight. Second, this paper adopts a feedback PID control system to adjust the fusion model weights. The control system comprises three parts: controller, control object, and measurement function. Among them, the control object is the optimized fusion model, and the measurement function is the image evaluation index. When designing the controller, this paper constructs a knowledge base based on the design experience of the existing fusion methods so that the controller can measure the difference between the evaluation result of the measurement function on the fused image and the target of the fusion task under the guidance of the knowledge base while adaptively adjusting the weights to meet the fusion task requirements. Therefore, this method can generate fusion results that meet the requirements without redesigning new fusion strategies for different tasks. Tests on multiple public datasets indicate the advantages of the proposed FusionIPCS over state-of-the-art methods.