针对增强现实在移动设备应用中特征点匹配速度缓慢的情况,提出了一种特征点匹配优化算法.该算法基于移动设备自带的重力感应系统,用重力方向代替原有的描述符方向,以此来降低传统ORB(oriented FASTand rotated BRIEF)算法中特征点方向角计算的时间复杂度.在实际移动设备上的测试结果表明,改进后的算法同时提高了匹配速度和匹配准确率,具有较好的效果.%Aiming at the problem that feature match on mobile devices was slow, this paper presented a new optimized algorithm which was based on gravity sensor. The algorithm replaced the original descriptor direction with the direction of gravity, so as to improve ORB( oriented FAST and rotated BRIEF). The experiments on mobile devices show that the method is faster and more precise on feature matching.
展开▼