基于连通区域特征的图像配准方法

摘要

提出了一种图像配准方法,对图像的平移、旋转以及缩放具有矫正效果.由于使用了连通区域作为配准对象,降低了算法复杂度,提高了效率.该方法对图像的连通区域进行检测,从而获得所有区域的属性特征,通过对这些特征进行多重筛选匹配得到图像变换前后的对应点,利用仿射变换矩阵结合插值运算进行图像的校正.实验结果表明,该方法对资源的占用比低,且具有较高的精度.%An image registration method is proposed, which could correct images' translation, rotation and scaling changes. The complexity of the algorithm is reduced due to the use of the connected regions as objects of registration, so that the algorithm becomes more efficient Firstly, the connected components of the image are detected, so that all the properties are acquired. Then matching the features by multiple filter and the corresponding points before and after image transformation are gotten. Finally, the image is corrected by the affine transformation matrix and interpolation. The experimental results show that the algorithm occupies less resource and has high accuracy.