光学学报, 2015, 35 (5): 0511003, 网络出版: 2015-04-08  

大面阵CMOS航空相机双重退化图像恢复

Restoration of Double Degraded Images in Large Frame CMOS Aerial Cameras
万磊 1,2,3,*张叶 1,3周达标 1,2,3贾平 1,3
作者单位
1 中国科学院长春光学精密机械与物理研究所, 吉林 长春 130033
2 中国科学院大学, 北京 100049
3 中国科学院航空光学成像与测量重点实验室, 吉林 长春 130033
摘要
由于采用电子卷帘式快门,大面阵互补金属氧化物半导体(CMOS)航空相机在高速成像时不仅会产生像移,还会引入由卷帘快门(RS)效应带来的几何畸变。传统的将两种图像退化因素分开进行处理的方法,过程繁复,计算复杂度高。针对航空前向飞行带来的线性像移和RS 畸变,通过分析各自的数学模型,推导出了二者之间的紧密联系,提出了基于线性像移参数估计的双重退化图像恢复方法。随后设计了航空前向飞行模拟实验,通过直线位移平台控制CMOS 相机以不同速度对目标景物成像。实验结果表明,RS 效应对像移参数的估计没有任何影响,最终恢复图像帧间变换保真度能提高到27 dB 左右。相比分步处理的方法,该方法在获得同样恢复效果的情况下,速度能提高3.5倍以上。
Abstract
Due to electronic rolling shutter, high-speed large frame complementary mental oxide semiconductor (CMOS) aerial cameras not only suffer from motion blur, but also lead to geometric distortion caused by rolling shutter effect. The conventional method which deals with these two degraded factors separately is very complicated and has a high computational complexity. To address the problems of linear motion and rolling shutter distortion resulting from aerial forward flight, a close relationship is derived between them through analyzing respective mathematical model. Furthermore, a method is proposed to restore double degraded images based on parameter estimation of linear motion. Then an experiment is designed to simulate aerial forward flight in which a CMOS camera controlled by the linear displacement stage imaging at different speed. Experiment results indicate that rolling shutter effect has no influence on the parameter estimation of linear motion. And the Interframe Transform Fidelity of restored images can be improved to about 27 dB. The calculation speed is 3.5 times faster than restoring step by step when obtaining the same good results.

万磊, 张叶, 周达标, 贾平. 大面阵CMOS航空相机双重退化图像恢复[J]. 光学学报, 2015, 35(5): 0511003. Wan Lei, Zhang Ye, Zhou Dabiao, Jia Ping. Restoration of Double Degraded Images in Large Frame CMOS Aerial Cameras[J]. Acta Optica Sinica, 2015, 35(5): 0511003.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!