摘要
介绍了一种波前解卷积中噪声抑制规整化的新方法,并将此方法应用于室内模拟点源实验中。该方法通过在图像复原算法中增加针对图像高频部分的限制条件来抑制高频噪声,以达到对图像复原问题病态特性的规整化。实验结果表明:该规整化方法可以有效地抑制解卷积过程中高频噪声的影响,恢复出达到理论衍射极限分辨率的图像。对于噪声水平较高的降质图像,通过这种解卷积方法可以有效地提高信噪比。同维纳逆滤波方法相比,该方法可以在有效抑制导致病态的高频噪声的基础上充分保持图像的低频;与基于贝叶斯估计的近视解卷积算法相比,该方法不需要知道噪声水平或噪声类型等先验知识,只是从噪声本质出发,通过抑制降质图像高频部分,有效地解决了病态特性问题。
This paper concerns solving deconvolution problem by constrain the noise in the degraded images in order to regularize the ill-posed characteristics. Adding constraint on degraded images during the deconvolution process, images of indoor point sources with aberration are restored. The results show that the method can constrain the noise in the high frequencies effectively and the restored images can reach the resolution of diffraction limit. Compared with the images restored by Wiener inverse filter, the images restored by noise constrained regularization have better vision quality,higher Strehl ratio and signal noise ratio.
出处
《强激光与粒子束》
EI
CAS
CSCD
北大核心
2007年第4期593-597,共5页
High Power Laser and Particle Beams
基金
国家863计划项目资助课题
关键词
图像复原
解卷积
噪声抑制
病态特性
规整化
Image restoration
Deconvolution
Noise constraint
Ill-posed characteristic
Regularization
