基于改进三角剖分插值EMD的多尺度边缘检测

为提高二维EMD分解速度,改善从本征模函数(IMF)图像提取边缘的质量,提出了一种改进三角剖分插值EMD的多尺度边缘检测算法.该算法首先通过邻域像素比较法得到图像极值点,利用改进的Delaunay三角剖分和三次样条插值函数进行曲面拟合,抑制了边界漏点问题,并用图像灰度均值改进了筛分停止准则,再对其分解得到的第一个IMF子图像进行小波多尺度分解提取图像边缘.通过仿真实验,结果表明该算法不仅能准确地提取图像边缘,还有效地抑制了噪声.仿真结果验证了该算法的可行性和有效性.

Multi-scale edge detection based on improved triangulation interpolation EMD

In order to improve the speed of two-dimensional empirical mode decomposition(2-D EMD)’ procedure and the quality of edges extract from intrinsic mode function(IMF) image,an algorithm based on triangulation interpolation two-dimensional empirical mode decomposition to detect image edges at multiple scales is proposed.This algorithm is realized using the method of neighborhood pixel comparison to detect regional maxima,Delaunay triangulation has been used to partition the seleeted extrema.then replaces the pixels that not contained in the Delaunay polygon through symmetry principle,which can restrain the variance phenomenon that appeared in the cubic spline interpolation,and improved the criterion of stopping sifting process using average gray of image,then extract edges from the first IMF image that was processed by multi-scale wavelet decomposition.Through the simulation experiment,results demonstrate that this method can not only detect image edge precisely,but also effectively restrain noise.Finally,the proposed algorithm is confirmed feasible and valid.