火力与指挥控制2026，Vol.51Issue(4)：83-90,8.DOI:10.3969/j.issn.1002-0640.2026.04.010
一种新的决策级融合目标检测算法
A Decision-level Fusion Object Detection Method for Visible and Infrared Images
摘要
Abstract
To address the issue of poor object detection performance in visible images under low-light conditions,a decision-level fusion object detection algorithm for infrared and visible images based on YOLOv11 and a multilayer perceptron(MLP)model is proposed.This algorithm introduces an MLP model in the detection result fusion stage,which dynamically adjusts fusion weights through data learning to adapt to variations in input distribution,thus effectively improving the adaptability of the algorithm.Experiments on the LLVIP dataset demonstrate that compared with existing mainstream methods,the proposed YOLOv11-MLP-based decision-level fusion algorithm achieves the best performance on both evaluation metrics AP50 and AP50-95,while maintaining real-time performance comparable to that of mainstream methods.The experimental results indicate that the YOLOv11-MLP-based decision-level algorithm can effectively fuse the complementary information of visible and infrared images and improve object detection performance under low-light conditions.关键词
可见光图像/红外图像/决策级融合/目标检测/深度学习/多层感知机Key words
visible image/infrared image/decision-level fusion/object detection/deep learning/MLP分类
信息技术与安全科学引用本文复制引用
虞亮亮,郝元宏,李秒..一种新的决策级融合目标检测算法[J].火力与指挥控制,2026,51(4):83-90,8.基金项目
装备预先研究基金资助项目(MKF20230069) （MKF20230069）
