摘要
为了提升计算机图形自适应匹配方法匹配效率、匹配准确度,实现计算机图形特征的精准匹配,该文提出基于SIFT算法的计算机图形自适应匹配方法。通过高斯卷积核构建尺度空间,在尺度空间引入DoG极值算子,提取计算机图形位于尺度空间的极值特征点,计算准确插值并以梯度幅值为基础,确定计算机图形极值特征点的主要方向,将特征点所在相邻区间分割成小区间,获取不同小区间的方位与向量特征描述因子,作为提取到的计算机图形极值特征点;自适应寻找最优化的距离比阈值,结合位置方向欧氏距离(POED)函数对所获取特征描述因子进行匹配,实现计算机图形的自适应匹配。实验表明,该方法可以实现对计算机图形自适应匹配,匹配用时较少,匹配准确度较高;对有噪声干扰的计算机图形依然可实现匹配,具有较高鲁棒性。
In order to improve the matching efficiency and accuracy of computer graphic adaptive matching methods,and achieve accurate matching of computer graphic features,this paper proposes a computer graphic adaptive matching method based on the SIFT algorithm.Construct a scale space using Gaussian convolution kernels,introduce the DoG extremum operator into the scale space,extract extreme feature points of computer graphics located in the scale space,calculate accurate interpolation based on gradient amplitude,determine the main direction of extreme feature points of computer graphics,divide adjacent intervals of feature points into small intervals,obtain orientation and vector feature description factors between different small intervals,and use them as extracted extreme feature points of computer graphics;Adaptive search for the optimal distance ratio threshold,combined with the position direction Euclidean distance(POED)function to match the obtained feature description factors,to achieve adaptive matching of computer graphics.The experiment shows that this method can achieve adaptive matching of computer graphics,with less matching time and higher matching accuracy;The computer graphics can still be matched against noise interference and have high robustness.
作者
郭林
沈东义
毛火明
GUO Lin;SHEN Dongyi;MAO Huoming(CNOOC(China)Limited Tianjin Branch,Tianjin 300459,China)
出处
《电子设计工程》
2025年第15期38-43,共6页
Electronic Design Engineering
