基于频域相位相关的自适应光学图像配准算法

ADAPTIVE OPTICAL IMAGES REGISTRATION ALGORITHM BASED ON FREQUENCY DOMAIN PHASE CORRELATION

自适应光学图像观测过程中,望远镜的机械抖动会使得各帧图像之间产生位移。直接对这些图像进行多帧复原,复原结果不佳。针对这一问题,结合自适应光学图像的特性,通过理论推导,提出了基于频域相位相关的自适应光学图像配准算法。算法通过计算配准图像间频域互功率谱并利用二维脉冲的位置解决自适应光学图像之间的配准问题。最后,利用仿真图像以及真实自适应光学图像进行验证,证明了算法的有效性,配准后的多帧复原图像质量有大幅提升。

Mechanical dithering of telescope can produce displacement between each frame of image sequence during the observing process of adaptive optical（ AO） image. The results of the multi-frame restoration algorithm directly using these AO images are very poor. Aiming at this problem,we proposed an AO images registration algorithm,which is based on frequency domain phase correlation,through theoretical derivation and combining the AO image characteristics. The algorithm achieves the registration of AO images by calculating the cross-power spectra of frequency domains between images and using the position of two-dimensional pulse. Finally,we verified the algorithm by using the simulation images and real AO images,the experiment demonstrated the effectiveness of the algorithm,and showed that the quality of the restored multi-frame image was greatly enhanced with the registration algorithm.