摘要
基于复数离散傅里叶变换(CDFT)频谱的共轭对称性,对传统的实数离散傅里叶变换(RDFT)做了改进,将改进后的RDFT算法用于维纳滤波器中来实现对运动模糊图像的恢复处理。一幅水平方向运动模糊,大小为512×512的8 bit灰度图像被用作测试图像,可以在19 ms内在Intel Pentium 4 3.0 GHz主频的PC机上用一维维纳滤波及改进的RDFT算法完成模糊恢复,满足实时处理的需要。实验结果显示,改进的RDFT能对CDFT的频谱数据实现有效压缩,节省一半的存储空间,使得维纳滤波等图像恢复频域算法的运算速度提高将近一倍;与CDFT相比,改进的RDFT算法没有丢失数据,能达到相同恢复精度。
Based on the conjugate symmetric frequency spectrum of Complex Discrete Fourier Transformation （CDFT） and the conventional Real Discrete Fourier Transformation（RDFT）, a modified RDFT is implemented in Wiener filter to restore aerial images degraded by motion or vibration. A horizontal rectilinear motion blurred image of 512×512 size and 8 bits as the test image in experiment can be restored by one dimensional Wiener filter and improved RDFT in 19 ms by an Intel Pentium 43.0 GHz CPU to meet real time requirement. Experiment results show that the modified RDFT efficiently reduce the data in the frequency spectrum of CDFT,which halves the storage space and nearly doubles the computational speed of the image restoration by Wiener filter. Meanwhile, compared with CDFT, the improved RDFT does not lose the data of frequency spectrum in the same restoration precision.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2007年第8期1287-1292,共6页
Optics and Precision Engineering
基金
国防科技预研基金资助项目(No.1040603)
关键词
航空成像
实数傅里叶变换
维纳滤波
图像恢复
aerial imaging
real discrete Fourier transformation
Wiener filter
image restoration
