两阶段平移、旋转图像高精度配准算法

图像配准在图像处理中是十分重要的,通常是许多现代图像处理和计算机视觉任务一个关键的预处理步骤,许多算法和技术已经被提出用来解决配准问题.本文提出一种稳健的两阶段层次配准算法,在频域组合相位相关和谱对消技术对平移和旋转图像获得亚像素精度的配准.算法第一阶段首先利用一维FFT技术实现图像的旋转以确定旋转角度,并利用这个旋转角度将两幅图像之间的运动简化为平移运动,然后利用相位相关法确定整数像素平移参数.算法第二阶段运用谱对消技术确定亚像素平移参数.提出的算法甚至在图像因为下采样而包含混淆误差情况下仍然能够获得亚像素精度配准.

Registration of Double Phases Translated and Rotated Image

The image registration takes an important part in image processing.It is a key preprocessing step in many modern image processing and computer vision tasks,and many algorithms and techniques have been proposed in order to resolve the registration problem.A robust double phases hierarchical image registration algorithm is presented which can combine the phase correlation with spectrum cancellation at the range of frequency to achieve high accuracy of subpixel through registration of translated and rotated images.In the first phase,the image rotation can be made using 1D FFT to determine the rotating angle and simplify the movement between images to translation,then to determine integral translated parameter with phase correlation.In the second phase,the spectrum cancellation is used to determine subpixel translated parameters.This algorithm can achieve subpixel accuracy registration even if the aliasing errors from undersampling exist in images.