基于灰度图像的兴趣点检测

Points-of-interest detection in the gray-value image

兴趣点是图像灰度在x ,y两个方向上都有很大变化的一类特征点 .对直接从灰度图像中检测兴趣点的方法进行研究 .鉴于以往算法检测出的兴趣点普遍存在着较大的点位置偏差 ,提出了一个基于Noble法的改进兴趣点检测算法 .实验表明 ,该算法很好地解决了Noble法的点位置偏差问题 .

The points of interest are those of useful low level features where the image gray value changes bi dimensionally. We study the points of interest detection methods directly from the gray value image. As there exists a severe point delocalization problem in the earlier approach, we present an improved points of interest detection algorithm based on the Nobles method. Good experimental results show that our algorithm can efficiently solve the point delocalization problem.