摘要
为了解决当前图像复原算法难以兼顾纹理与精细边缘的不足,提出了局部加权高斯-SAR联合先验模型的图像复原算法。引入局部自回归约束,利用高斯先验,构造局部加权高斯图像先验;并联合SAR先验,设计了高斯-联合先验模型,有效地防止过度平滑;并利用图像损坏模型与高斯-联合先验,建立其(Maximizing A Posteriori);基于最小优化技术,获取其下边界,将非凸问题转成凸问题,完成图像复原。对比测试结果显示:其算法的修复效果更佳,值最高,保留了丰富纹理与精细边缘;且复原图像的梯度分布与初始图像最接近。
In order to solve the difficulty of keeping balance between texture and fine edge with current image restoration algorithm,this paper proposed a new image restoration algorithm based on Combining Priori Model of Local Weighted Gauss-SAR.Firstly,the locally weighted GS image priori was constructed by introducing the Local autoregressive properties and TV function.The Gauss-SAR combining priori model was designed by combining SAR prior for preventing over-smooth of image texture,and the Maximizing A Posteriori was produced by image degradation model and GS-SAR combination priori.Simulation results showed that this algorithm had better reconstruction quality with higher PSNR value compared with current image restoration algorithm,and the texture and fine edge of restoration image was clearly visible.
出处
《微型电脑应用》
2015年第7期19-21,26,共4页
Microcomputer Applications
关键词
图像复原
局部自回归约束
高斯先验
联合先验模型
最小优化
梯度分布
Image Restoration
Local Autoregressive Constraint
GS Prior
Combining Priori Model
Minimization Optimization
Gradient Distribution
