The feature points are often used to describe the image for image matching. It has small amount of calculation and makes image matching work simply. The feature points tend to be relatively stable. It is difficult to change with the lighting conditions, image translation, rotation and scale. Image registration method based on feature points is currently a research hotspot. The principle of several mainstream feature point detection algorithm are introduced. And the performance of feature point detection algorithm is analyzed quantitatively. Feature point detection rate, accuracy rate, rotation invariance, and anti-noise to study the advantages and disadvantages of a variety of feature point detection algorithm are used. It can provide a good objective basis for selecting feature extraction algorithm based on the different quality of images.%在图像匹配中,使用特征点来描述图像,计算量小、匹配简单,而且特征点往往比较稳定,不易随光照条件和图像的平移、旋转和尺度等变化而变化.因此基于特征的图像配准方法是目前图像配准领域研究的热点.介绍了几种主流的特征点检测算法的原理,并对特征点检测算法性能进行了定量分析研究.从特征点检测速率、正确率、旋转不变性和抗噪性等方面进行了对比,研究了各种特征提取算法的优缺点.为根据不同图像质量选择合理的特征提取算法提供了良好的客观依据.
展开▼
