摘要
提出了一种基于核偏差估计的航天目标图像快速盲反卷积算法,基于自然图像的梯度具有长尾分布的特点,引入超拉普拉斯先验,针对模糊航天退化图像模糊核估计不精确的情况,在概率模型里添加一个关于点扩散函数的偏差估计用于补偿已知点扩散函数和真实的点扩散函数之间的误差,以增加算法的鲁棒性.该算法不仅减少了反卷积过程中时间复杂度高的问题,而且相较于其他快速算法更能保证图像恢复质量.该算法与目前图像去模糊方法相比具有较好的性能,并且与传统梯度先验模型在频域求解相比,能有效抑制阶梯效应衍生的截断条纹.
A fast deconvolution algorithm for space target image based on kernel deviation estimation was proposed.Based on the feature that the gradient of natural image had a long tail distribution,a hyper-Laplace prior was introduced.In the case of the kernel imprecision,a deviation estimation of the point diffusion function was added to the probability model to compensate the known point diffusion function and the real point diffusion function to increase the robustness of the algorithm.Because the added gradient prior will produce truncation effect in the process of solving in frequency domain,our algorithm can also effectively suppress the truncation fringe derived from the effect.Experimental results show that this algorithm can not only reduce the problem of high time complexity in the deconvolution process,but also ensure the image restoration quality compared with other fast algorithms,and has better performance than the current image deblurring method.
作者
华夏
徐卓帆
时愈
洪汉玉
HUA Xia;XU Zhuofan;SHI Yu;HONG Hanyu(Hubei Key Laboratory of Optical Information and Pattern Recognition,Wuhan Institute of Technology,Wuhan 430205,China;Hubei Research Centre of Video Image and High Denition Projection,Wuhan Institute of Technology,Wuhan 430205,China;School of Electrical and Information Engineering,Wuhan Institute of Technology,Wuhan 430205,China)
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2020年第3期57-62,共6页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(61433007,61671337,61701353,61801337)
湖北省教育厅科学技术研究资助项目(Q20171510).
关键词
核偏差估计
航天目标图像
超拉普拉斯
截断效应
鲁棒性
快速盲反卷积
kernel error estimate
astronautic target image
hyper-Laplace
truncation effect
robustness
fast deconvolution
作者简介
华夏(1987-),男,讲师;通信作者:时愈,副教授,E-mail:shiyu0125@163.com.
