Written by: Dodge(Lang HE) asdsay@gmail.com
Updated date: 2023-11-16
Variational Mode Decomposition for Python in 2D
VMD, aka Variational Mode Decomposition, is a signal processing tool that decompse the input signal into different band-limited IMFs.
VMD_2D, means we are processing 2D signal (Two dimension should usually have same length). Project VMD_2D_Python is an imitation of that in MATLAB. Spectrum-based decomposition of a 2D input signal into k band-separated modes.
In this project, I used a grey picture to test. The function VMD2D only needs Numpy, but we also need OpenCV and matplotlib to read and print the picture.
Please also pay attention to line 34-35 in VMD2D.py :
# Maximum number of iterations
N = 3000
In the original Matlab code, it was a solid 3000. However under my test, the sample pictgure does not converge. Luckily, for this picture, it is practically the same if I set N = 100. So feel free to change the N value.
If you are looking for document to describe Variational mode decomposition, please turn to the original paper:
K. Dragomiretskiy, D. Zosso, Variational Mode Decomposition, IEEE Trans. on Signal Processing (in press) please check here for update reference: http://dx.doi.org/10.1109/TSP.2013.2288675
作者：Dodge asdsay@gmail.com 更新日期：2023-11-16
VMD（变分模态分解）是一种信号处理算法，可以将输入信号分解为不同带限的内禀模态函数（IMFs）。 VMD_2D意味着我们正在处理二维信号（通常两个维度应该长度相同）。项目是MATLAB中实现的模仿。基于频谱的二维输入信号分解为k个带分离模式。 本项目VMD_2D_Python是参考于其在MATLAB中的实现。基于频谱的二维输入信号分解为k个带分离模式。
在这个项目中，我用一张灰度图片进行测试。本项目VMD_2D_Python仅需要Numpy，但我们还需要OpenCV和matplotlib两个库来读取和显示图片。
还请注意 VMD2D.py第34-35行：
# Maximum number of iterations
N = 3000
在原始的Matlab代码中，N是固定值3000。然而在测试中，样本图片计算的误差没有收敛。发现对于这张图片，如果设置N = 100，实际效果几乎就收敛了。请用户更改N值测试效果。
如果需要描述变分模态分解的文档，可参阅原始论文： K. Dragomiretskiy, D. Zosso, Variational Mode Decomposition, IEEE Trans. on Signal Processing (in press) http://dx.doi.org/10.1109/TSP.2013.2288675