particle swarm optimization

[imago.git] / hough.py

diff --git a/hough.py b/hough.py
index 7ac3aac..ec8058c 100644 (file)
--- a/hough.py
+++ b/hough.py
@@ -1,43 +1,49 @@
-from PIL import Image
 from math import sin, cos, pi
-from commons import clear
-
-def transform(image):
-
-    image_l = image.load()
-    size = image.size
-       
-    dt = pi / size[1]
-    initial_angle = (pi / 4) + (dt / 2)
-
-    matrix = [[0]*size[1] for _ in xrange(size[0])]
 
-    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
+from PIL import Image
 
-    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
+import pcf
+
+class Hough:
+    def __init__(self, size):
+        self.size = size
+        self.dt = pi / size[1]
+        self.initial_angle = (pi / 4) + (self.dt / 2)
+
+    def transform(self, image):
+        image_s = pcf.hough(image.size, image.tostring(), self.initial_angle, self.dt)
+        image = Image.fromstring('L', image.size, image_s)
+        return image
+
+    def lines_from_list(self, p_list):
+        lines = []
+        for p in p_list:
+            lines.append(self.angle_distance(p))
+        return lines
+
+    def all_lines_h(self, image):
+        im_l = image.load()
+        lines1 = []
+        for x in xrange(image.size[0] / 2):
+            for y in xrange(image.size[1]):
+                if im_l[x, y]:
+                    lines1.append(self.angle_distance((x, y)))
+        lines2 = []
+        for x in xrange(image.size[0] / 2, image.size[0]):
+            for y in xrange(image.size[1]):
+                if im_l[x, y]:
+                    lines2.append(self.angle_distance((x, y)))
+        return [lines1, lines2]
+
+    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 angle_distance(self, point):
+        return (self.dt * point[1] + self.initial_angle, point[0] - self.size[0] / 2)
+