In order to solve the low accuracy of common target recognition results in polarized dark light scenes,this paper proposes a fu-sion algorithm based on convolutional neural network of polarization degree images and visible light images,designs a new loss function to form an unsupervised learning process,introduces Laplace operator to improve the quality of fused image,and finally the polarization information of the target to be measured is effectively combined with the visible light information;A fused target detection algorithm is proposed based on the improved YOLOv5 algorithm for the target detection of the fused target,the coordinate attention(CA)attention mechanism of the network framework is added to combine the channel attention mechanism with the spatial attention mechanism.The proposed network is trained and tested on a homemade dataset,the results show that the fused image achieves better visual effects subjectively and objectively,compared with the optimal YOLOv5s model,the precision and recall rate of the improved YOLOv5 algorithm reach 89.3％and 82.5％,respectively,and the mean average precisions increase by 2.6％and 1.8％,respectively.
维普
万方数据