摘要
传统的凸集投影(projection onto convex sets,POCS)算法只能得到低对比度、低信噪比、边缘模糊的图像。为了解决上述问题,提出了1种改进的基于视觉机制的POCS超分辨率重建算法,引入可变校正阈值的数据一致性约束,以突出目标边缘和滤除背景噪声;对传统POCS算法的插值算法进行改进,采用基于梯度的插值算法进行图像的初始估计。实验结果表明,改进的POCS算法可以获得高质量的图像对比度,另外其重建效率、结构相似性指数和峰值信噪比也都有所改进。
The traditional projection onto convex sets（POCS）algorithm can only images with low contrast,low signal to noise ratio,and edge blurring.In order to solve the above problems,an improved POCS super-resolution reconstruction algorithm based on visual mechanism is proposed,which introduces the data consistency constraint of variable correction threshold to highlight the target edge and filter out background noise.The interpolation algorithm of the traditional POCS algorithm is also improved,and the initial estimation of the image is carried out by using the gradient interpolation algorithm.The experimental results show that the improved POCS algorithm can obtain high quality image contrast,and its reconstruction efficiency,structural similarity index and peak signal to noise ratio are also improved.
出处
《中国科技论文》
北大核心
2017年第14期1655-1658,1684,共5页
China Sciencepaper
基金
山西省科技攻关项目(20130321007-02)
关键词
凸集投影
超分辨率重建
梯度插值
可变阈值
convex projection
super resolution reconstruction
gradient interpolation
variable threshold
