Add remove_period_patterns.py

datasets/raise/fft/remove_period_patterns.py

@@ -0,0 +1,54 @@
+import numpy as np
+from astropy.convolution import Gaussian2DKernel, interpolate_replace_nans
+from PIL import Image
+import os
+import sys
+from scipy import fftpack
+
+sys.path.insert(0, '../')
+
+from utils import Color, mergeSingleColorChannelImagesAccordingToBayerFilter
+import matplotlib.pyplot as plt
+
+PREFIX = 'mean_rafael_230424_mean_'
+X_STDDEV = 2
+
+def getImageByColor(color):
+    filePath = PREFIX + f'{color}.npy'
+    image = np.load(filePath)
+    return image
+
+singleColorChannelImages = {color: getImageByColor(color) for color in Color}
+multipleColorsImage = mergeSingleColorChannelImagesAccordingToBayerFilter(singleColorChannelImages)
+image = multipleColorsImage
+
+fft1 = abs(fftpack.fft2(image))
+originalFft1 = fft1.copy()
+
+# This example is intended to demonstrate how astropy.convolve and
+# scipy.convolve handle missing data, so we start by setting the brightest
+# pixels to NaN to simulate a "saturated" data set
+height, width = fft1.shape
+for x in range(width):
+    fft1[height // 2, x] = np.nan
+
+middleX = width // 2
+RANGE = 1
+for verticalLineX in [width // 4, width // 2, round(3 * width / 4)]:
+    for y in range(height):
+        for x in range(verticalLineX - RANGE, verticalLineX + RANGE + 1):
+            fft1[y, x] = np.nan
+
+# We smooth with a Gaussian kernel
+kernel = Gaussian2DKernel(x_stddev = X_STDDEV)
+
+# create a "fixed" image with NaNs replaced by interpolated values
+fixedImage = interpolate_replace_nans(fft1, kernel)
+
+figure, axes = plt.subplots(1, 2, sharex = True, sharey = True)
+
+axes[0].imshow(np.log10(originalFft1))
+axes[1].imshow(np.log10(fixedImage))
+#plt.imshow(np.log10(fixedImage))
+plt.show()
+#plt.imsave('fft.png', np.log10(fixedImage))