摘要： 光纤耦合图像之间存在亚像元级位移问题，为此，提出一种基于图像互补信息的超分辨率重建算法。采用传统相位相关法进行整像元级配准，通过拟合多个谱峰值进行亚像元级配准，利用待插值点周围的4条B样条曲线实现图像插值融合。实验结果表明，该算法能提高图像空间分辨率，且复杂度较低、重建图像质量较好。

关键词: 传像光纤束, 亚像元, 相位相关, 多峰拟合, B样条插值, 超分辨率重建

Abstract: As the image coupled by the adjacent layer fibers has a sub-pixel dislocation, a super-reconstruction algorithm based on image complementary information is proposed. It makes pixel-level registration by traditional phase correlation, and sub-pixel registration by multi-peak fitting. Four B-spline curves for ultimate interpolation are used to fuse the matched image with the reference image. Experimental results show that the proposed algorithm can effectively raise the spatial resolution with less complexity, and the reconstruction image owns better quality.

Key words: image-carrying fiber bundle, sub-pixel, phase correlation, multi-peak fitting, B-spline interpolation, super-resolution reconstruction

中图分类号: 

• TN911.73