摘要
FAST特征提取算法阈值选取固定,因此不能满足不同图像的特征点抽取要求,并且提取的结果存在着多个特征点块的现象。针对这些缺陷,首先采用动态全局阈值对原始灰度图像进行初步提取得到候选特征点,然后采取动态局部阈值和非极大值抑制法进一步对候选特征点进行筛选,从而达到自适应选取阈值和抑制多个特征点块的目的。实验表明,改进后的算法稳定性高,对不同光照和对比度情况下有一定的适应能力,并且运算量相对比于其他一些特征提取算法要小得多,满足实时应用的要求。
There are still some problems with The FAST (Features from Accelerated Segment Test) Feature Detection. The fixed threshold selection of FAST Algorithm can't meet the requirements of feature extraction with different images. Moreo- ver, there are a large number of feature points gathering together in the feature extraction results. To solve these problems, the dynamic threshold and non-maxima suppression method are proposed to make futher choosing candidate feature points, so as to achieve setting self-adaptive threshold and inhibiting the feature point blocks. Our experimental results show that the proposed algorithm can reduce the amount of calculation and has high stability and adaptability in different conditions of illu- mination and contrast which can satisfy the real time requirement.
出处
《指挥控制与仿真》
2013年第2期47-53,共7页
Command Control & Simulation
关键词
特征点提取
聚集率
动态阈值
非极大值抑制
feature detection
aggregation rate
self-adaptive threshold
non-maximal suppression