python - 一维数组的 Richardson-Lucy 反卷积

Question

我正在寻找适用于一维数组(例如光谱数据)的 Richardson-Lucy 反卷积算法的实现。我试过scikit-image ，但显然它只适用于图像。

Answer 1

您是否尝试过在单行/单列二维数组上使用restoration.richardson_lucy？它是否按预期工作？

这是一个基于 http://scikit-image.org/docs/dev/auto_examples/filters/plot_deconvolution.html 的示例(参见输入单元格 3 和 4):

In [1]: import numpy as np
   ...: import matplotlib.pyplot as plt
   ...: 
   ...: from scipy.signal import convolve2d as conv2
   ...: 
   ...: from skimage import color, data, restoration
   ...: 
   ...: astro = color.rgb2gray(data.astronaut())
   ...: 

In [2]: 
   ...: psf = np.ones((5, 5)) / 25
   ...: astro = conv2(astro, psf, 'same')
   ...: # Add Noise to Image
   ...: astro_noisy = astro.copy()
   ...: astro_noisy += (np.random.poisson(lam=25, size=astro.shape) - 10) / 255.
   ...: 
   ...: 

In [3]: astro_1d = astro[:1, :]
In [4]: psf_1d = psf[:1, :] * 5

In [5]: deconvolved_RL = restoration.richardson_lucy(astro_1d, psf_1d, iteration
   ...: s=30)
   ...: 
   ...: 

In [8]: deconvolved_RL[0][:10]
Out[8]: 
array([  3.68349589e-06,   4.64232976e-03,   8.96492325e-01,
         2.92227252e-01,   2.27669473e-01,   1.63909318e-01,
         2.62231088e-01,   5.63304220e-01,   4.29589937e-01,
         3.21857292e-01])

In [9]: astro_1d[0][:10]
Out[9]: 
array([ 0.20156543,  0.25178911,  0.31006612,  0.29581576,  0.30208733,
        0.32490093,  0.35101666,  0.36213184,  0.35174074,  0.318339  ])

如果您发现转换为 2D 确实不方便，请随时在 GitHub 上提出问题。