X-Git-Url: http://git.tomasm.cz/imago.git/blobdiff_plain/a24a89991790610c9bb8754b269b376350d807e3..ba7978f0ac03e56190df7ed04f8a6e03c4193546:/hough.py?ds=sidebyside

diff --git a/hough.py b/hough.py
index 7ac3aac..d01416a 100644
--- a/hough.py
+++ b/hough.py
@@ -1,43 +1,93 @@
-from PIL import Image
 from math import sin, cos, pi
+
+from PIL import Image
+
 from commons import clear
 
-def transform(image):
+class Hough:
+    def __init__(self, size):
+        self.size = size
+        self.dt = pi / size[1]
+        self.initial_angle = (pi / 4) + (self.dt / 2)
 
-    image_l = image.load()
-    size = image.size
-       
-    dt = pi / size[1]
-    initial_angle = (pi / 4) + (dt / 2)
+    def transform(self, image):
+        image_l = image.load()
+        size = image.size
+        
+        matrix = [[0]*size[1] for _ in xrange(size[0])]
 
-    matrix = [[0]*size[1] for _ in xrange(size[0])]
+        dt = self.dt
+        initial_angle = self.initial_angle
 
-    for x in xrange(size[0]):
-        clear()
-        print "hough transform: {0}/{1}".format(x + 1, size[0])
-        for y in xrange(size[1]):
-            if image_l[x, y]:
-                # for every angle:
-                for a in xrange(size[1]):
-                    # find the distance:
-                    # distance is the dot product of vector (x, y) - centerpoint
-                    # and a unit vector orthogonal to the angle
-                    distance = (((x - (size[0] / 2)) * sin((dt * a) + initial_angle)) + 
-                                ((y - (size[1] / 2)) * -cos((dt * a) + initial_angle)) +
-                                size[0] / 2)
-                    column = int(round(distance)) # column of the matrix closest to the distance
-                    if column >= 0 and column < size[0]:
-                        matrix[column][a] += 1
-
-    new_image = Image.new('L', size)
-    new_image_l = new_image.load()
-
-    minimum = min([min(m) for m in matrix])
-
-    maximum = max([max(m) for m in matrix]) - minimum
-
-    for y in xrange(size[1]):
         for x in xrange(size[0]):
-            new_image_l[x, y] = (float(matrix[x][y] - minimum) / maximum) * 255
+            clear()
+            print "hough transform: {0:>3}/{1}".format(x + 1, size[0])
+            for y in xrange(size[1]):
+                if image_l[x, y]:
+                    # for every angle:
+                    for a in xrange(size[1]):
+                        # find the distance:
+                        # distance is the dot product of vector (x, y) - centerpoint
+                        # and a unit vector orthogonal to the angle
+                        distance = (((x - (size[0] / 2)) * sin((dt * a) + initial_angle)) + 
+                                    ((y - (size[1] / 2)) * -cos((dt * a) + initial_angle)) +
+                                    size[0] / 2)
+                        # column of the matrix closest to the distance
+                        column = int(round(distance)) 
+                        if column >= 0 and column < size[0]:
+                            matrix[column][a] += 1
+
+        new_image = Image.new('L', size)
+        new_image_l = new_image.load()
+
+        minimum = min([min(m) for m in matrix])
+
+        maximum = max([max(m) for m in matrix]) - minimum
+
+        for y in xrange(size[1]):
+            for x in xrange(size[0]):
+                new_image_l[x, y] = (float(matrix[x][y] - minimum) / maximum) * 255
             
-    return new_image
+        return new_image
+
+    def all_lines(self, image):
+        im_l = image.load()
+        lines = []
+        for x in xrange(image.size[0]):
+            for y in xrange(image.size[1]):
+                if im_l[x, y]:
+                    lines.append(self.angle_distance((x, y)))
+        return lines
+    
+    def find_angle_distance(self, image):
+        image_l = image.load()
+
+        points = []
+
+        count = 0
+        point_x = 0
+        point_y = 0
+        for x in xrange(image.size[0] / 2):
+            for y in xrange(image.size[1] / 2, image.size[1]):
+                if image_l[x, y]:
+                    count += 1
+                    point_x += x
+                    point_y += y
+        points.append((float(point_x) / count, float(point_y) / count))
+
+        count = 0
+        point_x = 0
+        point_y = 0
+        for x in xrange(image.size[0] / 2, image.size[0]):
+            for y in xrange(image.size[1] / 2, image.size[1]):
+                if image_l[x, y]:
+                    count += 1
+                    point_x += x
+                    point_y += y
+        points.append((float(point_x) / count, float(point_y) / count))
+
+        return [self.angle_distance(p) for p in points]
+
+    def angle_distance(self, point):
+        return (self.dt * point[1] + self.initial_angle, point[0] - self.size[0] / 2)
+