改进SIFT算法的图像双向配准技术

Image bidirectional registration technology based on improved SIFT algorithm

为提高图像配准的效率和准确度,满足多种应用场景的适用性,文章提出一种基于k-d树的双向配准算法。该算法运用准欧氏距离进行匹配,运算简单,在原图像与待配准图像中双向搜索相匹配的特征点,图像配准的准确率高。实验结果表明,改进的算法准确率提高了6%,时间缩短了31%。

To improve the efficiency and accuracy of image registration and adapt to various application scenarios,a bidirectional registration algorithm based on K-D tree is proposed.The algorithm uses quasi-Euclidean distance to match images,and the calculation is simple.A bidirectional search is conducted for matching feature points between the original image and the image to be registered with high accuracy.Experimental results show that the improved algorithm improves the accuracy by 6%and shortens the time by 31%.