一种基于边缘特征的亚像素投影配准算法

为了提高红外图像配准准确度,充分利用图像的边缘信息,结合线性插值,对原有的投影算法引入新的图像相似性评价函数——均方误差函数(Mean Sequare Erron,MSE),提出了基于边缘特征的亚像素投影配准算法.提取图像边缘特征明显的部分做配准能提高匹配概率,引入积分投影能有效降低算法的计算复杂度.算法中还运用了线性插值,使配准准确度达到亚像素级水平.

Edge-feature-based Sub-pixel Registration Algorithm Using Integral Projections

To enhance the accuracy of the IR image registration,an edge-feature-based registration algorithm using integral projection was proposed.It was based on edge feature of an image and introduces mean square error function as a new similarity measure function different from the original projection algorithm.The algorithm increased the matching probability by extracting the image block abundant in edge and reduces the computation complexity using the projection.A linear interpolation was also employed to ensure that the algorithm has a sub-pixel precision.