一种基于SURF的图像配准改进算法

为了更好地构造逼真的虚拟环境，提高虚拟场景中图像配准的效率，提出了一种改进的SURF算法。改进算法根据特征点的数量和其间疏密关系作为判定条件，可以在更短时间内得到数量适当且分布相对均匀的图像特征点，同时在特征点匹配阶段利用Hessian矩阵迹的正负性提高特征点匹配的速度。针对误匹配影响图像拼接准确性的问题，采用随机采样算法（RANSAC）提高匹配的精确度。实验结果表明该算法节省了特征点检测和匹配的时间，提高了匹配效率。

Improved algorithm of image registration based on SURF

To better construct a virtual environment and increase the image registration efficiency in the virtual scene, the speed up the robust features (SURF) algorithm was improved. The algorithm was based on the number of feature points and density relationship which determine the conditions, The number of the appropriate and relatively uniform distribution of the feature points can achieve within a shorter time. The positive and negative quantity of Hessian matrix was used to increase the speed of the feature point matching.To solve the problem which wrong matches influence the precision of image mosaic, random sample consensus (RANSAC) was used to improve the matching accuracy. The experimental results show that the algorithm can save the feature point detection and the matching of time, and improve the matching efficiency. The proposed algorithm of image registration is more effective than the SURF image registration method.