基于改进ResNet算法的太阳黑子分类方法研究  

Research on sunspot classification method based on improved ResNet algorithm

作　　者：樊石鸣 FAN Shiming(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650500

出　　处：《现代电子技术》2024年第5期80-84,共5页Modern Electronics Technique

摘　　要：文中提出一种基于改进的ResNet的太阳黑子图像分类方法,该方法在ResNet的基础上引入了空洞卷积和残差连接等技术,增强了训练模型的特征提取能力以及感受野。在模型训练过程中,采用交叉熵损失函数和SGD优化器进行参数优化,以提高模型的准确性和泛化能力。其中深度可分离卷积被用于替代传统的卷积操作,以减少模型的参数量和计算量。最后,在太阳黑子图像分类的公共数据集上进行实验评估。实验结果表明,提出的基于改进的ResNet模型在太阳黑子图像分类任务上表现出较高的准确率和鲁棒性,相比于传统的ResNet模型,可以取得更好的分类效果。该方法为太阳黑子图像分类提供了一种新的思路和技术方案,对于太阳活动预测和环境监测等领域具有重要意义。A sunspot image classification method based on improved ResNet is proposed in this study.In the method,techniques such as dilated convolution and residual concatenation are introduced on the basis of ResNet to enhance the feature extraction ability of the training model and the perceptual field.In the process of model training,the cross⁃entropy loss function and SGD(stochastic gradient descent)optimizer are used for parameter optimization,so as to improve the accuracy and generalization ability of the model.Among them,depth⁃separable convolution is used to replace the traditional convolution operation to reduce the number of parameters and the computational effort of the model.An experimental evaluation is performed on a public dataset for sunspot image classification.The experimental results show that the model based on improved ResNet exhibits high accuracy and robustness in the task of sunspot image classification and can achieve better classification results in comparison with the traditional ResNet model.The proposed method provides a new idea and technical scheme for sunspot image classification,which is important in the fields of solar activity prediction and environmental monitoring.

关 键 词：太阳黑子 ResNet 空洞卷积 残差连接 交叉熵 SGD 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP399[电子电信—信息与通信工程]