摘要
针对传统的凸集投影(POCS)超分辨率图像重建算法经常出现的边缘模糊问题,在传统POCS算法原理基础上,文章使用基于梯度插值的算法生成POCS重建的初始估计图像,然后对中心点为边缘像素的空间点扩散函数(PSF)进行修正,使其系数沿梯度大的方向减小,梯度小的方向保持不变。利用改进的POCS算法进行了超分辨率图像重建实验,结果表明,图像峰值信噪比由原来的27.29dB达到28.12dB。该方法有效地保持了边缘特性,改善了超分辨率图像重建质量。
The traditional projection onto the convex sets super-resolution image reconstruction algorithm leads to the blur edge of reconstructed high resolution image.Based on the traditional algorithm,a method based on gradient interpolation is proposed,which is used to generate the initial estimation of the super-resolution image.Furthermore,by modifying the point spread function（PSF）,the center of which is on the edge,the coefficients along the direction of big gradient are reduced,while others keep constant.We reconstruct a set of low resolution images with the updating algorithm.The experimental results show that the edges kept very well,and the performance of the reconstructed images is improved.
出处
《液晶与显示》
CAS
CSCD
北大核心
2010年第6期873-878,共6页
Chinese Journal of Liquid Crystals and Displays
基金
"973"国家自然基金项目(No.2009CB72400105)
关键词
超分辨率
POCS算法
空间点扩散函数
边缘保持
CANNY
航拍图像
super-resolution
projection to convex sets （POCS）
point spread function（PSF）
edge constraint
Canny
aerial photography
