Color channels upscaled by bicubic interpolation

For color images, we previously applied our RAISR algorithm only to the luminance channel, whereas in the Google's RAISR implementation and by others, the color channels are upsampled by bicubic interpolation. We can also use bicubic interpolation, but this will not affect our reported PSNR results as it is dependent on luminance only. Based on comments in issue 1, in this commit, we are switching to applying bicubic interpolation on the chromatic channels.

Author: Cejo Konuparamban Lonappan

.gitignore

@@ -1,5 +1,6 @@
 # Byte-compiled / optimized / DLL files
 __pycache__/
+.idea/
 *.py[cod]
 *$py.class
 

Filter/Qfactor_coh2

@@ -28,7 +28,6 @@ def BGR2YCbCr(im):
     mat = np.array([[24.966, 128.553, 65.481],[112, -74.203, -37.797], [-18.214, -93.786, 112]])
     mat = mat.T
     offset = np.array([[[16, 128, 128]]])
-
     if im.dtype == 'uint8':
         mat = mat/255
         out = np.dot(im,mat) + offset
@@ -48,7 +47,6 @@ def YCbCr2BGR(im):
     mat = mat.T
     mat = np.linalg.inv(mat)
     offset = np.array([[[16, 128, 128]]])
-
     if im.dtype == 'uint8':
         mat = mat * 255
         out = np.dot((im - offset),mat)
@@ -64,12 +62,12 @@ def YCbCr2BGR(im):
     return out
 
 def im2double(im):
-    min_val = np.min(im.ravel())
-    max_val = np.max(im.ravel())
     if im.dtype == 'uint8':
         out = im.astype('float') / 255
     elif im.dtype == 'uint16':
         out = im.astype('float') / 65535
+    elif im.dtype == 'float':
+        out = im
     else:
         assert False
     out = np.clip(out, 0, 1)
@@ -123,7 +121,6 @@ def is_greyimage(im):
 
 @nb.jit(nopython=True, parallel=True)
 def Grad(patchX,patchY,weight):
-
     gx = patchX.ravel()
     gy = patchY.ravel()
     G = np.vstack((gx,gy)).T
@@ -133,17 +130,13 @@ def Grad(patchX,patchY,weight):
     index= w.argsort()[::-1]
     w = w[index]
     v = v[:,index]
-
     lamda = w[0]
-
     u = (np.sqrt(w[0]) - np.sqrt(w[1]))/(np.sqrt(w[0]) + np.sqrt(w[1]) + 0.00000000000000001)
-
     return lamda,u
 
 @nb.jit(nopython=True, parallel=True)
 def HashTable(patchX,patchY,weight, Qangle,Qstrength,Qcoherence,stre,cohe):
     assert (len(stre)== Qstrength-1) and (len(cohe)==Qcoherence-1),"Quantization number should be equal"
-
     gx = patchX.ravel()
     gy = patchY.ravel()
     G = np.vstack((gx,gy)).T
@@ -153,28 +146,20 @@ def HashTable(patchX,patchY,weight, Qangle,Qstrength,Qcoherence,stre,cohe):
     index= w.argsort()[::-1]
     w = w[index]
     v = v[:,index]
-
     theta = atan2(v[1,0], v[0,0])
     if theta<0:
         theta = theta+pi
     theta = floor(theta/(pi/Qangle))
-
     lamda = w[0]
-
     u = (np.sqrt(w[0]) - np.sqrt(w[1]))/(np.sqrt(w[0]) + np.sqrt(w[1]) + 0.00000000000000001)
-
     if isnan(u):
         u=1
-
     if theta>Qangle-1:
         theta = Qangle-1
     if theta<0:
         theta = 0
-
     lamda = np.searchsorted(stre,lamda)
-
     u = np.searchsorted(cohe,u)
-
     return theta,lamda,u
 
 @nb.jit(nopython=True, parallel=True)
@@ -233,7 +218,6 @@ def Blending2(LR, HR):
 def Backprojection(LR, HR, maxIter):
     H, W = LR.shape
     H1, W1 = HR.shape
-
     w = Gaussian2d((5,5), 10)
     w = w**2
     w = w/sum(np.ravel(w))
@@ -242,7 +226,6 @@ def Backprojection(LR, HR, maxIter):
         imd = LR - im_L
         im_d = imresize(imd, (H1, W1), interp='bicubic', mode='F')
         HR = HR + convolve2d(im_d, w, 'same')
-
     return HR
 
 def createFolder(directory):

Filter/Qfactor_coh3

@@ -63,7 +63,7 @@ Google's RAISR implementation was trained on 10000 advertising banner images. Ou
 <p align="center">
   <img src="Comparison_result.png">
 </p>
-A high resolution image after downscaling it by 3 was used as the input low resolution image to both bicubic interpolation and our implementation of RAISR algorithm. A comparison of the resulted output image for this testcase with bicubic interpolation and our implementation of RAISR is shown in the figure above. 
+A high resolution image after downscaling it by 2 was used as the input low resolution image to both bicubic interpolation and our implementation of RAISR algorithm. A comparison of the resulted output image for this testcase with bicubic interpolation and our implementation of RAISR is shown in the figure above.
 
 ## Citations
 Please cite the following publications if you plan to use the code or the results for your research: 

Filter/Qfactor_coh4

@@ -51,13 +51,13 @@
 Qstrength = 3  # Quantization factor of strength =3
 Qcoherence = 3  # Quantization factor of coherence =3
 
-with open("Filter/filter"+str(R), "rb") as fp:
+with open("Filters/filter"+str(R), "rb") as fp:
     h = pickle.load(fp)
 
-with open("Filter/Qfactor_str"+str(R), "rb") as sp:
+with open("Filters/Qfactor_str"+str(R), "rb") as sp:
     stre = pickle.load(sp)
 
-with open("Filter/Qfactor_coh"+str(R), "rb") as cp:
+with open("Filters/Qfactor_coh"+str(R), "rb") as cp:
     cohe = pickle.load(cp)
 
 filelist = make_dataset(testPath)
@@ -168,8 +168,9 @@ def TestProcess(i):
 
     if len(im_uint8.shape) > 2:
         result_ycbcr = np.zeros((H, W, 3))
-        result_ycbcr[:, :, 1:3] = im_ycbcr[:, :, 1:3]
         result_ycbcr[:, :, 0] = im_blending * 255
+        result_ycbcr[:, :, 1] = Prepare(im_ycbcr[:, :, 1], patchSize, R) * 255
+        result_ycbcr[:, :, 2] = Prepare(im_ycbcr[:, :, 2], patchSize, R) * 255
         result_ycbcr = result_ycbcr[region].astype('uint8')
         result_RAISR = YCbCr2BGR(result_ycbcr)
     else: