Apply Context-Adaptive Interpolator

algorithms/image_utils/image_utils.py

@@ -7,4 +7,7 @@ def randomGaussianImage(scale, size):
 
 def showImageWithMatplotlib(npArray):
     plt.imshow(npArray)
-    plt.show()
+    plt.show()
+
+def toPilImage(npArray):
+    return Image.fromarray(npArray)

datasets/fake/generate_dataset.py

@@ -15,7 +15,7 @@ from context_adaptive_interpolator import contextAdaptiveInterpolator
 
 sys.path.insert(0, '../../algorithms/image_utils/')
 
-from image_utils import showImageWithMatplotlib, randomGaussianImage
+from image_utils import showImageWithMatplotlib, randomGaussianImage, toPilImage
 from tqdm import tqdm
 
 IMAGE_SIZE = 64
@@ -37,9 +37,6 @@ imagesWithPrnu = [[imageWithoutPrnu + prnus[phoneIndex] for imageWithoutPrnu in
 
 allImages = np.max([np.max(imagesWithoutPrnu) + np.max(prnus) + np.max(imagesWithPrnu)])
 
-def toPilImage(npArray):
-    return Image.fromarray(npArray)
-
 def showImageWithPil(npArray):
     npArray -= npArray.min()
     npArray = (npArray / npArray.max()) * 255

datasets/noise_free_test_images/estimate_prnu.py

@@ -7,12 +7,16 @@ import sys
 
 sys.path.insert(0, '../../algorithms/image_utils/')
 
-from image_utils import showImageWithMatplotlib, randomGaussianImage
+from image_utils import showImageWithMatplotlib, randomGaussianImage, toPilImage
+
+sys.path.insert(0, '../../algorithms/context_adaptive_interpolator/')
+
+from context_adaptive_interpolator import contextAdaptiveInterpolator
 
 datasetPath = 'no_noise_images'
 # Note that contrarily to `datasets/fake/`, here we do not have images being Gaussian with `scale` `1` but actual images with pixel values between 0 and 255.
 # In addition to the range difference, note that the distribution in the first set of images was a Gaussian and here is very different and specific.
-PRNU_FACTOR = 0.1
+PRNU_FACTOR = 0.15
 IMAGE_SIZE_SHAPE = (469, 704)
 
 np.random.seed(0)
@@ -21,11 +25,22 @@ np.random.seed(0)
 prnuPil = Image.open('prnu.png').convert('F')
 prnuNpArray = np.array(prnuPil) * PRNU_FACTOR
 
+def isIn256Range(x):
+    return 0 <= x and x <= 255
+
 for imageName in os.listdir(datasetPath):
     if imageName.endswith('.png'):
         imagePath = f'{datasetPath}/{imageName}'
-        imagePil = Image.open(imagePath).convert('F')
-        imageNpArray = np.array(imagePil)
-        #showImageWithMatplotlib(imageNpArray)
-        showImageWithMatplotlib(imageNpArray + prnuNpArray)
-        break
+        imageWithoutPrnuPil = Image.open(imagePath).convert('F')
+        imageWithoutPrnuNpArray = np.array(imageWithoutPrnuPil)
+        #showImageWithMatplotlib(imageWithoutPrnuNpArray)
+        imageWithPrnuNpArray = imageWithoutPrnuNpArray + prnuNpArray
+        showImageWithMatplotlib(imageWithPrnuNpArray)
+        break
+        assert all([isIn256Range(extreme) for extreme in [imageWithPrnuNpArray.max(), imageWithPrnuNpArray.min()]]), 'Adding the PRNU resulted in out of 256 bounds image'
+        imageWithPrnuPil = toPilImage(imageWithPrnuNpArray)
+        imageWithPrnuCaiPil = contextAdaptiveInterpolator(imageWithPrnuPil.load(), imageWithPrnuPil)
+        imageWithPrnuCaiNpArray = np.array(imageWithPrnuCaiPil)
+        showImageWithMatplotlib(imageWithPrnuCaiNpArray)
+        break
+