基于小波变换的图像配准

Image Registration Based on Wavelet Transform

图像配准是信息融合处理中的重要环节。本文分析了图像配准的数学模型,并对小波变换进行了研究。基于小波理论,提出了一种高精度的图像配准方法。该方法利用小波变换将图像分成若干层次,按照互信息最大的原则对小波分解各层的近似分量求取其配准参数,最后通过迭代实现图像配准。实验结果表明,该方法配准精度高、可靠性好,较之传统的方法有明显的优越性。

Image registration is one of the crucial steps in the fusion processing of image. In this paper, mathematics model of image registration is analyzed, and wavelet transform theory is researched. An image registration method based on wavelet is proposed. Decomposed wavelet sub-bands of the original images are obtained using the wavelet transform. Determine the optimal registration parameters of decomposed wavelet sub-bands in the light of the maximum mutual information criterion, and registration is realized using iterative algorithm finally. The results show that the method has some advantages such as higher precision and reliability. And this method outperforms conventional approaches.