摘要
在图像匹配过程中,针对传统归一化积相关(NCC)算法计算量大的问题,提出一种对NCC进行改进的图像匹配快速算法。该算法首先使用差分求和定理改造NCC相似度量函数,以降低匹配计算量。然后提出模板区域分割,设定阈值,进一步去除大量不必要的计算,优化匹配搜索过程,实现了快速匹配。实验结果证明,与传统的匹配算法相比,在保证精度的前提下,计算复杂度大大降低。
In the image matching process, the computational complexity was the major problem of the traditional NCC （normalized product correlation） algorithm. An image matching fast algorithm for improving the NCC is presented. First, it combined with the summation theorem of difference to improve the NCC for reducing the amount of calculation of matching. Second, it showed region segmentation for the template and then set threshold to get rid of some calculation to achieve rapid matching. The experimental results prove that the computational complexity is reduced greatly compared with the traditional matching algorithm by the premise of the precision
出处
《现代电子技术》
2010年第22期107-109,共3页
Modern Electronics Technique
基金
国家自然科学基金资助项目(60672135)
关键词
图像匹配
归一化积相关
相似度函数
区域分割
差分
image matching
normalized product correlation
similarity function
region segmentation
difference
