Restore other than raw images support

datasets/raise/extract_noise.py

@@ -21,8 +21,8 @@ imagesFolderPathFileName = imagesFolderPath.replace('/', '_')
 denoiser = 'mean'
 
 raiseNotFlatFields = False
-# `[Color.RED, Color.GREEN_RIGHT, ...]` or `Color`.
-colors = Color
+# `[Color.RED, Color.GREEN_RIGHT, ...]` or `Color` or `[None]` for not raw images.
+colors = [None]
 
 if denoiser != 'mean':
     denoise = getattr(skimage.restoration, f'denoise_{denoiser}')
@@ -80,28 +80,30 @@ def getImageNpArray(imageFileName, computeExtremes, color):
 
             match color:
                 case Color.RED:
-                    colorRawImageVisible = redRawImageVisible
+                    imageNpArray = redRawImageVisible
                 case Color.GREEN_RIGHT:
-                    colorRawImageVisible = greenRightRawImageVisible
+                    imageNpArray = greenRightRawImageVisible
                 case Color.GREEN_BOTTOM:
-                    colorRawImageVisible = greenBottomRawImageVisible
+                    imageNpArray = greenBottomRawImageVisible
                 case Color.BLUE:
-                    colorRawImageVisible = blueRawImageVisible
-
-            if computeExtremes:
-                colorRawImageVisibleMin = colorRawImageVisible.min()
-                colorRawImageVisibleMax = colorRawImageVisible.max()
-                if minColor is None or colorRawImageVisibleMin < minColor:
-                    minColor = colorRawImageVisibleMin
-                if maxColor is None or colorRawImageVisibleMax > maxColor:
-                    maxColor = colorRawImageVisibleMax
-                return
-            imageNpArray = (colorRawImageVisible - minColor) / (maxColor - minColor)
-            # Pay attention to range of values expected by the denoiser.
-            # Indeed if provide the thousands valued raw image, then the denoiser only returns values between 0 and 1 and making the difference between both look pointless.
+                    imageNpArray = blueRawImageVisible
     else:
         imagePil = Image.open(imageFilePath)
         imageNpArray = img_as_float(np.array(imagePil))
+
+    if computeExtremes:
+        colorRawImageVisibleMin = imageNpArray.min()
+        colorRawImageVisibleMax = imageNpArray.max()
+        if minColor is None or colorRawImageVisibleMin < minColor:
+            minColor = colorRawImageVisibleMin
+        if maxColor is None or colorRawImageVisibleMax > maxColor:
+            maxColor = colorRawImageVisibleMax
+        return
+
+    if imageFileName.endswith('.NEF') or imageFileName.endswith('.ARW'):
+        imageNpArray = (imageNpArray - minColor) / (maxColor - minColor)
+        # Pay attention to range of values expected by the denoiser.
+        # Indeed if provide the thousands valued raw image, then the denoiser only returns values between 0 and 1 and making the difference between both look pointless.
     return imageNpArray
 
 # `color` is the actual color to estimate PRNU with.
@@ -143,10 +145,10 @@ def saveNpArray(fileName, npArray):
 
 if denoiser == 'mean':
     means = {}
-    for color in Color:
+    for color in colors:
         colorMean = None
         numberOfImagesInColorMean = 0
-        for imageFileName in tqdm(imagesFileNames, f'Computing mean of {color} images'):
+        for imageFileName in tqdm(imagesFileNames, f'Computing mean of {color colored images'):
             imageNpArray = getImageNpArray(imageFileName, False, color)
             if colorMean is None:
                 colorMean = imageNpArray

datasets/raise/show_mean_noise.py

@@ -3,6 +3,8 @@ import matplotlib.pyplot as plt
 
 fileName = 'mean_flat-field_nef_wavelet_blue'
 npArray = np.load(f'{fileName}.npy')
+# For other than raw images:
+#npArray = (npArray - npArray.min()) / (npArray.max() - npArray.min())
 plt.imsave(f'{fileName}.png', npArray)
 #plt.imshow(npArray)
 #plt.show()