摘要
构建了一种新的局部梯度方向直方图,同时定义了特征点的主方向,从而提出了一种具有旋转不变性的图像配准算法。首先采用高斯加权求模技术对特征点邻域内的像素的梯度作直方图统计,确定出具有旋转不变性和抗噪性的特征点主方向;然后用主方向作角度直方图统计,确定待配准图像之间的旋转角度。根据得到的特征点信息及旋转角度定义了特征点对互信息匹配准则,这样使得新配准算法对于图像间旋转角度的范围没有限制,获得了良好的配准效果。
A novel gradient orientation histogram was built using a Gaussian-weighted circular window and a major orientation to each feature point was assigned based on local image properties. Then the orientation difference between two target images was calculated from an angle histogram, which was presented with feature point major orientations instead of former gradient orientations. Based on the rotation angle and feature points, the mutual information of feature points was defined to obtain match pairs. Finally, a novel registration algorithm was obtained that was invariance to rotation and robust to noise. The experiments show that this approach has no restriction to the rotation angle between the images and the registration results are satisfying.
出处
《计算机应用研究》
CSCD
北大核心
2007年第3期312-314,共3页
Application Research of Computers
基金
重庆市自然科学基金资助项目(CSTC2005BA2002)
关键词
图像配准
主方向
梯度方向直方图
旋转不变性
互信息
image registration
major orientation
gradient orientation histogram
rotation invariance
mutual information
