UAV swarms have crucial applications in modern military, geological exploration, and 5G/6G communication fields. As communication nodes, drones frequently exchange wireless data with other drones, and data privacy protection is currently one of the most urgent research topics. Based on this, this article proposes an efficient isomorphic federated learning algorithm for unmanned aerial vehicle clusters. First, set the adaptive loss differential adaptive parameter, with the initial value set to a floating point number not greater than 0.01, then activate it with the Tanh function, participate in the training as part of the loss function during the training process, and optimize it by gradient descent. When the UAV node uploads the gradient to the wingman, sum the adaptive loss differential parameter with the gradient matrix as a disturbance term. Then, based on the dynamic confidence matrix, high-quality drones are selected to participate in the gradient security aggregation of drones, achieving the selection of high-quality drones. The built-in global neural network of the drone transmits shared parameters indiscriminately to each drone through broadcast to achieve updates to the shared parameters. The comparative experiments of our algorithm on the Fashion and Cifar10 datasets show that our algorithm has higher accuracy, with the highest accuracy improvement of 4.42% on the Mnist dataset and 8.22% on the Cifar10 dataset.
We propose an integral infrared scene simulation system. The proposed system, which is based on the parameters of the thermal physical property and optical property, computes the radiation distribution of the scenery on the focus plane of the camera according to the scene of the geometrical parameter, the position and intensity of the light source, the location and direction of the camera and so on. Then the radiation distribution is mapped to the space of gray, and we finally obtain the virtual image of the scene. The proposed system includes eight modules namely basic data maintaining, model importing, scene saving, geometry parameters setting and infrared property parameters of the scene, data pre-processing, infrared scene simulation, and scene loading. The proposed system organizes all the data by the mode of database lookup table that stores all relative parameters and computation results of different states to avoid repetitive computation. Experimental results show that the proposed system produces three dimension infrared images in real time to some extent, and can reach 60 frames/second in simple scene drawing and 20 frames/second in complex scene drawing. Experimental results also show that the simulated images can represent infrared features of the scenery to a certain degree.
In real traffic scenes, the quality of captured images are generally low due to some factors such as lighting conditions, and occlusion on. All of these factors are challengeable for automated recognition algorithms of traffic signs. Deep learning has provided a new way to solve this kind of problems recently. The deep network can automatically learn features from a large number of data samples and obtain an excellent recognition performance. We therefore approach this task of recognition of traffic signs as a general vision problem, with few assumptions related to road signs. We propose a model of Convolutional Neural Network (CNN) and apply the model to the task of traffic signs recognition. The proposed model adopts deep CNN as the supervised learning model, directly takes the collected traffic signs image as the input, alternates the convolutional layer and subsampling layer, and automatically extracts the features for the recognition of the traffic signs images. The proposed model includes an input layer, three convolutional layers, three subsampling layers, a fully-connected layer, and an output layer. To validate the proposed model, the experiments are implemented using the public dataset of China competition of fuzzy image processing. Experimental results show that the proposed model produces a recognition accuracy of 99.01 % on the training dataset, and yield a record of 92% on the preliminary contest within the fourth best.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.