摘要
直接波前解卷积可以有效地克服大气湍流对天文观测的影响,但解卷积问题具有病态特性,必须进行规整化。基于奇异值分解的规整化方法在处理图像边缘等高频部分不够理想;而WVD方法仅适用某些解卷积问题。在提出一种基于小波包变换的新规整化方法的基础上,将此方法应用于室内模拟点源实验中,并与基于奇异值分解规整化的维纳逆滤波进行了对比,实验结果表明:该规整化方法可以有效地解决解卷积问题的病态特性,应用该规整化方法所恢复的图像质量明显提高。
Deconvolution from wavefront sensor is a powerful tool for compensating the atmosphere turbulence. With ill- posed characteristic,the classic regularized methods based on singular value decomposition e. g. , the wiener inverse filter method can not get preferable results at higher frequencies of the image; On the other hand, the WVD method is only suitable for some specific deconvolution problems. This paper concerns solving deconvolution problem by a new regularization with the help of discrete wavelet packet. Images of point sources indoors with random aberration are restored. The restored images are compared with those restored by. The results show that this method can solve the ill-posed problem effectively and the images restored by this method get better quality than those restored by wiener inverse filter.
出处
《强激光与粒子束》
EI
CAS
CSCD
北大核心
2008年第3期377-382,共6页
High Power Laser and Particle Beams
基金
国家高技术发展计划项目
中国科学院“优秀博士毕业论文、院长奖获得者科研启动专项奖金”资助课题
关键词
图像复原
解卷积
小波包
病态特性
规整化
Image restoration
Deconvolution
Wavelet packet
Ill-posed characteristic
Regularization
