基于二维EMD的Harris角点检测算法

为解决Harris角点检测算法在多尺度条件下无法正确提取角点的问题,本文将经验模式分解(EMD)方法运用到二维图像特征点提取中。先利用二维EMD方法将图像分解到多个图像细节层,并定义为本征模函数(IMF),再利用Harris算子对各图像细节层进行角点检测,最后采用层层筛选的方法提取角点。对比实验结果表明,新算法得到的角点更加丰富,抗噪性增强,明显提高了图像角点检测性能。

Harris Corner Detection Algorithm Based on Two-dimensional EMD

To solve the problem of the Harris corner detection algorithm which cannot detect corners in multi-scales, this paper applies the Empirical Mode Decomposition (EMD) to corner detection of two-dimensional images. First, the image is decomposed into some image detail layers (Intrinsic Mode Functions, IMF) by the EMD, then use Harris corner detection operator on the details of the layer. Finally, filter the corner points layer by layer. The contrast experimental results show that, the new algorithm gets more corners, enhances the noise immunity and improves the image corner detection performance significantly.