Research on Camouflage Target Detection Method Based on Dual Band Optics and SAR Image Fusion

In order to counteract the camouflage of targets, this paper trains an improved YOLOv7 deep learning target detection framework using visible light images, infrared images, and SAR images from the same perspective as the dataset. Propose an improved D-S evidence theory fusion method based on Jaccard coefficients. This article introduces the entropy of the power set as the multiplication factor when calculating the Jaccard coefficient matrix, which improves the performance of evaluating conflicts between multiple input evidence and enhances the effectiveness and adaptability of D-S evidence theory in dealing with evidence with significant conflicts. Propose an improved deep learning object detection algorithm based on YOLOv7. At the input end of the backbone network, three types of images were simultaneously input, and the convolutional structure was optimized and attention mechanism was introduced; The improved D-S evidence theory fusion formula is introduced into the Loss function, and it is responsible for the evidence fusion of the decision vectors output from the full connection layer in the network structure to determine the final detection results. Verified on the dataset collected in this article, compared to using the original YOLOv7 for single source image camouflage target detection, the improved target detection algorithm has better detection accuracy when facing camouflage targets and can meet the requirements of real-time detection.