光学学报, 2019, 39 (12): 1228001, 网络出版: 2019-12-06
综合深度卷积神经网络的摆扫影像反演恢复算法 
下载： 943次
Inversion Restoring Algorithm for Whiskbroom Scanning Images Synthesized with Deep Convolutional Neural Network
图 & 表
图 3. 垂轨方向上星下点地物的像元数目计算示意图
Fig. 3. Calculation of pixel number of nadir ground scene perpendicular to track
图 4. 垂轨方向上原畸变图与拉伸图像的像元对应示意图
Fig. 4. Schematic of corresponding pixels of distorted and extended images perpendicular to track
图 6. 沿轨方向上原畸变图与拉伸图像的像元对应示意图。(a)原畸变图;(b)拉伸图像
Fig. 6. Schematics of corresponding pixels of distorted and extended images along track. (a) Original distorted image; (b) extended image
图 9. 卫星摆扫成像实验装置示意图
Fig. 9. Schematic of experimental device of satellite whiskbroom scanning imaging
图 10. 摆扫影像的畸变校正结果。(a)地面真值;(b)摆扫影像模拟图;(c)摆扫影像恢复图
Fig. 10. Distortion correction result of whiskbroom scanning image. (a) Ground truth; (b) simulated whiskbroom scanning image; (c) distortion-corrected whiskbroom scanning image
图 11. 摆扫线性靶标的畸变校正结果。(a)摆扫线性靶标模拟图;(b)摆扫线性靶标恢复图
Fig. 11. Distortion correction result of whiskbroom scanning linear target. (a) Simulated whiskbroom scanning linear target; (b) distortion-corrected whiskbroom scanning linear target
图 12. 畸变校正后摆扫影像的增强结果。(a)畸变校正后的摆扫影像;(b) Bicubic结果;(c) SRCNN结果;(d) wbi-SRCNN结果。左图为wall,右图为roof
Fig. 12. Enhanced results of distortion-corrected whiskbroom scanning images. (a) Distortion-corrected whiskbroom scanning images; (b) result of Bicubic; (c) result of SRCNN; (d) result of wbi-SRCNN. Left is wall and right is roof
表 1在轨成像与地面模拟参数
Table1. Parameters for on-orbit imaging and ground simulation
|
表 2测试图像的NR-IQA结果
Table2. NR-IQA results of test images
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
徐超, 金光, 杨秀彬, 徐婷婷, 常琳. 综合深度卷积神经网络的摆扫影像反演恢复算法[J]. 光学学报, 2019, 39(12): 1228001. Chao Xu, Guang Jin, Xiubin Yang, Tingting Xu, Lin Chang. Inversion Restoring Algorithm for Whiskbroom Scanning Images Synthesized with Deep Convolutional Neural Network[J]. Acta Optica Sinica, 2019, 39(12): 1228001.
PDF全文
