accelerate

depth/depth_render/hwclient.py

@@ -9,9 +9,11 @@ import numpy as np
 import PIL
 from PIL import Image
 import scipy.misc
-
+import os
 from cube2equi import find_corresponding_pixel
 
+from transfer import transfer2
+
 
 context = zmq.Context()
 
@@ -23,7 +25,7 @@ socket = context.socket(zmq.REQ)
 socket.connect("tcp://localhost:5555")
 
 #  Do 10 requests, waiting each time for a response
-for request in range(10):
+for request in range(6):
     print("Sending request %s ..." % request)
     socket.send(b"Hello")
 
@@ -45,11 +47,29 @@ for request in range(10):
     print("Received reply %s [ %s ]" % (request, data))
 
 
-
+def transfer(in_img, coords, h, w):
+    out_img = np.zeros((h,w,3)).astype(np.uint8)
+    
+    for ycoord in range(0, h):
+        for xcoord in range(0, w):
+            ind, corrx, corry = coords[ycoord, xcoord, :]   
+            out_img[ycoord, xcoord, :] =  in_img[ind, corrx, corry, :]
+    return out_img
+    
 def convert_img():
     #inimg = Image.open(infile)
 
     inimg = InImg()
+    
+    in_imgs =  np.array([np.array(Image.open("0.tiff")),
+    np.array(Image.open("1.tiff")),
+    np.array(Image.open("2.tiff")),
+    np.array(Image.open("3.tiff")),
+    np.array(Image.open("4.tiff")),
+    np.array(Image.open("5.tiff"))]).astype(np.uint8)
+    
+    print(in_imgs.shape)
+    
     wo, ho = inimg.grid * 4, inimg.grid * 3
 
     # Calculate height and width of output image, and size of each square face
@@ -58,34 +78,36 @@ def convert_img():
     n = ho/3
 
     # Create new image with width w, and height h
-    outimg = Image.new('RGB', (w, h))
+    outimg = np.zeros((h,w,3)).astype(np.uint8)
 
+    if not os.path.isfile('coord.npy'):
+        coords = np.zeros((h,w,3)).astype(np.int32)
+        
+        for ycoord in range(0, h):
+            for xcoord in range(0, w):
+                corrx, corry = find_corresponding_pixel(xcoord, ycoord, w, h, n)
+                coords[ycoord, xcoord, :] = inimg.getpixel((corrx, corry))
+        
+        np.save('coord.npy', coords)
+    else:
+        coords = np.load('coord.npy')
+    
+    
     # For each pixel in output image find colour value from input image
-    for ycoord in range(0, h):
-        for xcoord in range(0, w):
-            corrx, corry = find_corresponding_pixel(xcoord, ycoord, w, h, n)
-
-            outimg.putpixel((xcoord, ycoord), inimg.getpixel((corrx, corry)))
-        # Print progress percentage
-        print str(round((float(ycoord)/float(h))*100, 2)) + '%'
-
+    outimg = transfer2(in_imgs, coords, h, w)
+    
+    
 
+    outimg = Image.fromarray(outimg)
     outimg.save('output', 'PNG')
 
 
 class InImg(object):
     def __init__(self):
         self.grid = 512
-        self.img0 = Image.open("0.tiff")
-        self.img1 = Image.open("1.tiff")
-        self.img2 = Image.open("2.tiff")
-        self.img3 = Image.open("3.tiff")
-        self.img4 = Image.open("4.tiff")
-        self.img5 = Image.open("5.tiff")
 
     def getpixel(self, key):
         corrx, corry = key[0], key[1]
-        imgs = [[self.img0], [self.img1, self.img2, self.img3, self.img4], [self.img5]]
         
         indx = int(corrx / self.grid)
         indy = int(corry / self.grid)
@@ -93,9 +115,11 @@ class InImg(object):
         remx = int(corrx % self.grid)
         remy = int(corry % self.grid)
 
-        if (indy == 0  or indy == 2):
-            return imgs[indy][0].getpixel((remx, remy))
+        if (indy == 0):
+            return (0, remx, remy)
+        elif (indy == 2):
+            return (5, remx, remy)
         else:
-            return imgs[1][indx].getpixel((remx, remy))
+            return (indx + 1, remx, remy)
 
-convert_img()
+convert_img()

depth/depth_render/setup.py

@@ -0,0 +1,4 @@
+from distutils.core import setup
+from Cython.Build import cythonize
+
+setup(name="transfer", ext_modules=cythonize('transfer.pyx'),)

depth/depth_render/transfer.pyx

@@ -0,0 +1,19 @@
+import numpy as np
+
+def transfer2(unsigned char [:,:,:,:] in_img, int [:,:,:]coords, int h, int w):
+    
+    cdef unsigned char [:,:,:] out_img = np.zeros((h,w,3)).astype(np.uint8)
+    
+    cdef int xcoord, ycoord
+    cdef int ind, corrx, corry
+    
+    cdef int c
+    
+    for ycoord in range(0, h):
+        for xcoord in range(0, w):
+            ind = coords[ycoord, xcoord, 0]   
+            corrx = coords[ycoord, xcoord, 1]
+            corry = coords[ycoord, xcoord, 2]  
+            for c in range(3):
+                out_img[ycoord, xcoord, c] =  in_img[ind, corrx, corry, c]
+    return np.array(out_img)