基于改进SIFT的室内全景图像配准算法研究

Research on indoor panoramic image registration algorithm based on improved SIFT

图像配准技术是计算机视觉领域的重要基础,SIFT算法作为高效的特征提取描述算法在图像配准领域有着广泛应用。为了解决传统SIFT算法在图像特征点提取中性能较低的问题,提出了一种改进的SIFT图像特征匹配算法。该改进算法在原始SIFT算法的基础上,优化了SIFT算法中构建高斯金字塔的流程,并采用BRIEF特征描述子替换SIFT特征描述子。实验结果证明,该算法与原始SIFT相比,配准时间缩短为原来时间的2/3、配准准确率提高了10多个百分点,可以高效完成室内全景图像特征匹配工作。

Image registration technology is an important foundation in the field of computer vision.As an efficient feature extraction and description algorithm,the SIFT algorithm has been widely used in the field of image registration.In order to solve the problem that traditional SIFT algorithm is inefficient in image feature point extraction,an improved SIFT image feature matching algorithm is proposed.Based on the original SIFT algorithm,the improved algorithm optimizes the process of building a Gaussian pyramid in the SIFT algorithm,and replaces the SIFT feature descriptor with the BRIEF feature descriptor.The results show that compared with the original SIFT,the algorithm has shorter registration time and higher registration accuracy,registration time is shortened to 2/3 of the original time,and the registration accuracy is increased by more than 10 percentage points,can efficiently complete the indoor panoramic image feature matching work.