激光与光电子学进展, 2020, 57 (20): 201018, 网络出版: 2020-10-13
基于多尺度生成对抗网络的SAR图像样本增广 下载: 1170次
Data Augmentation in SAR Images Based on Multi-Scale Generative Adversarial Networks
图 & 表
图 4. 线性卷积层与1×1卷积层。(a)线性卷积层;(b) 1×1卷积层
Fig. 4. Linear convolutional layer and 1×1 convolutional layer. (a) Linear convolution layer; (b) 1×1 convolution layer
图 7. 由单一图像生成的图像。(a)小尺寸舰船的图像1;(b)小尺寸舰船的图像2;(c)背景带噪声的图像;(d)大尺寸舰船的图像
Fig. 7. Image generated from single image. (a) Image1 of small ship; (b) image2 of small ship; (c) image with noise in background; (d) image of large ship
图 8. 不同网络生成的图像。(a)用于训练的图像;(b)原网络生成的图像;(c)改进后网络生成的图像
Fig. 8. Images generated by different networks. (a) Images used for training; (b) images generated by the original network; (c) images generated by the improved network
图 9. 不同数据集训练的错检情况。(a)正确检测结果;(b) SSDD数据集;(c) SSDD数据集+生成的样本数据集
Fig. 9. Error detection of training on different data sets. (a) Correct test results; (b) SSDD data set; (c) SSDD data set + generated sample data set
图 10. 不同数据集训练的虚警情况。(a)正确检测结果;(b) SSDD数据集;(c) SSDD数据集+生成的样本数据集
Fig. 10. False alarms of training on different data sets. (a) Correct test results; (b) SSDD data set; (c) SSDD data set + generated sample data set
图 11. 不同数据集训练的漏检情况。(a)正确检测结果;(b) SSDD数据集;(c) SSDD数据集+生成的样本数据集
Fig. 11. Missed detection of training on different data sets. (a) Correct test results; (b) SSDD data set; (c) SSDD data set + generated sample data set
表 1生成器的参数
Table1. Parameters of the generator
|
表 2不同方法生成图像的AP
Table2. AP of different methods to generate images
|
李诗怡, 付光远, 崔忠马, 杨小婷, 汪洪桥, 陈雨魁. 基于多尺度生成对抗网络的SAR图像样本增广[J]. 激光与光电子学进展, 2020, 57(20): 201018. Shiyi Li, Guangyuan Fu, Zhongma Cui, Xiaoting Yang, Hongqiao Wang, Yukui Chen. Data Augmentation in SAR Images Based on Multi-Scale Generative Adversarial Networks[J]. Laser & Optoelectronics Progress, 2020, 57(20): 201018.