利用尺度空间下特征点进行匹配的电子稳像方法

在电子稳像过程中，获取准确的图像运动矢量是稳像的关键，而尺度不变特征转换（SIFT）算法可以较准确地提取运动矢量。为此给出了一种基于尺度不变特征变换的特征提取和匹配的电子稳像方法。SIFT算法是一种在不同尺度空间下提取特征点的方法，该方法首先进行尺度空间极值点检测，然后对特征点定位，最后进行特征向量生成与匹配。实验结果表明，该方法具有多量性，提取特征点数较多且特征匹配点对具有较高的准确率，可以获取较理想的稳像效果。

Image stabilization using matching feature points in scale space

In the process of electronic image stabilization, getting correct image motion vector is the key to image stabilization algorithm, Scale Invariant Feature Transform (SIFT) algorithm can accurately extract the motion vector. This paper present a image stabilization algorithm based on the extraction and tracking of Scale Invariant Feature Transform (SIFT).This algorithm is a method to extract feature points in different scales space. SIFT algorithm in the first detect extrema points in scale space, then feature points location, the final steps are generation of feature vectors and feature vectors matching. The experiment results show that this algorithm with large amounts of nature can extract a number of feature points, but also has high accuracy. It can get a better image stabilization effect.