CAJ | 학술논문

针对模糊退化PS F未知的图像超分辨率重建问题，提出了一种以学习重建结果为参考的模糊图像PS F辨识与重建方法。对学习重建算法获得的退化图像的恢复结果，采用不同尺度的方向滤波提取恢复效果良好的强边缘区域构建参考信息块，引导最大后验概率框架下的超分辨率重建算法按照正确的PSF重建高分辨率图像。以散焦模糊图像为例进行超分辨率实验。实验结果表明，该算法能够准确辨识PS F ，提升了图像重建的质量。
침대모호퇴화PS F미지적도상초분변솔중건문제，제출료일충이학습중건결과위삼고적모호도상PS F변식여중건방법。대학습중건산법획득적퇴화도상적회복결과，채용불동척도적방향려파제취회복효과량호적강변연구역구건삼고신식괴，인도최대후험개솔광가하적초분변솔중건산법안조정학적PSF중건고분변솔도상。이산초모호도상위례진행초분변솔실험。실험결과표명，해산법능구준학변식PS F ，제승료도상중건적질량。
For super‐resolution of image with unknown PSF ,a method to PSF identification and super resolution using the result of learning‐based algorithm as a reference is proposed .Recovered from the learning‐based algorithm ,the strong edges of good performances are selected as reference blocks through filters with different scales and different directions .With the help of these blocks ,the MAP algorithm will be guaranteed to construct the high resolution image by the optimal PSF .Super‐resolution recon‐struction experiments show this method turns out a good PSF and a good resolution image .