一种改进的多尺度Harris特征点检测方法

An Improved Multi-scale Harris Feature Point Detection Method

多尺度Harris方法检查到的特征点存在很多冗余点,虽然Harris-Laplace方法可以除去一些冗余点,但是还会出现一个局部结构内存在多个特征点的情况或一个特征点代表多个不同尺度的局部结构。为此,提出一种改进的方法,在检测多尺度Harris特征点时进行跟踪分组,使代表同一个局部结构的特征点被分为一组,用归一化的Laplace函数去除冗余点,再利用点的度量值选出最能代表该局部结构的特征点。实验结果证明,该方法能够有效去除冗余点,在模糊和旋转变换时性能优于Harris-Laplace方法,具有尺度不变的特性。

There are many redundant feature points by multi-scale Harris method to check. Although Harris-Laplace method can remove some redundant points, there still are more than one feature points in a local structure or a few different scale local structures represented a feature points. This paper present an improved method of checking Harris-Laplace feature points, that track and group it while checking feature points so that some feature points that represent the same local structure are divided into a group, then use Laplace function to remove redundant feature points and the comer measure to select the most representative of the local structure of the comer. It is proved by experiments that the method can effectively remove redundant feature points. While fuzzy and rotation transforming, the method is better than Harris-Laplace, and has a scale-invariant features.