Abstract

The BRISK, AKAZE, and ORB detectors of singular points are compared. In a series of experiments on a set of 10 000 images, the performance characteristics are estimated for algorithms of search for singular points. These characteristics include the fraction of correctly found positions of singular points, the fraction of correct matches, and the number of matches at comparison of two arbitrarily selected images. The dependence of the average number of false matches on the value of admissible similarity of descriptors is calculated. The distributions of the number of correct and false matches are plotted.

中文翻译：

---

图像奇异点检测器的比较及其统计特征的估计

摘要

比较了奇异点的 BRISK、AKAZE 和 ORB 检测器。在一组 10 000 张图像上的一系列实验中，估计了搜索奇异点算法的性能特征。这些特征包括正确找到奇异点位置的比例、正确匹配的比例以及在比较两个任意选择的图像时匹配的数量。计算错误匹配的平均数量对描述符的可容许相似度值的依赖性。绘制了正确和错误匹配数的分布。