Merge branch 'master' of github.com:taddeus/licenseplates

docs/report.tex

@@ -356,19 +356,23 @@ essential that you use the correct folder and subfolder naming scheme. The schem
 is as follows:
 
 \begin{enumerate}
-
-\item A main folder called 'images' placed in the current directory as the src folder.
-\item In the images folder there have to be three folders. Images, Infos and LearningSet.
-\item The Images and Infos folder contain subfolders which are numbered ($0001$ to possibly $9999$).
-\item In each of the subfolders the data (i.e the images or xml files) can be placed.
-And have to be named $00991_XXXXX.ext$, where XXXXX can be $00000 to 99999$.
-\item For loops in the script currently only go up to 9 subfolders, with a maximum
-of containing 100 images or xml files. These numbers have to be adjusted if the scripts 
-are being used, but with a bigger dataset.
+    \item A main folder called `images' placed in the current directory as the
+    src folder.
+    \item In the images folder there have to be three folders.  Images, Infos
+    and LearningSet.
+    \item The Images and Infos folder contain subfolders which are numbered
+    ($0001$ to possibly $9999$).
+    \item In each of the subfolders the data (i.e the images or xml files) can
+    be placed.  And have to be named $00991_XXXXX.ext$, where XXXXX can be
+    $00000 to 99999$.
+    \item For-loops in the script currently only go up to 9 subfolders, with a
+    maximum of containing 100 images or xml files. These numbers have to be
+    adjusted if the scripts are being used, but with a bigger dataset.
 \end{enumerate}
 
-It is ofcourse possible to use your own naming scheme. A search for the $filename$ variable will most
-likely find the occurences where the naming scheme is implemented.
+It is of course possible to use your own naming scheme. A search for the
+$filename$ variable will most likely find the occurences where the naming
+scheme is implemented.
 
 
 \subsection*{\texttt{create\_characters.py}}
@@ -385,16 +389,19 @@ likely find the occurences where the naming scheme is implemented.
 
 \subsection*{\texttt{generate\_learning\_set.py}}
 
-Usage of this script could be minimal, since you only need to extract the letters carefully
-and succesfully once. Then other scripts in this list can use the extracted images. Most
-likely the other scripts will use caching to speed up the system to. But in short, the script will create images of a single character based on a given dataset
-of license plate images and corresponding xml files. If the xml files give correct
-locations of the characters they can be extracted. The workhorse of this script is
-$plate = xml_to_LicensePlate(filename, save_character=1)$. Where save_character is
-an optional variable. If set it will save the image in the LearningSet folder and
-pick the correct subfolder based on the character value. So if the XML says a character
-is an 'A' it will be placed in the 'A' folder. These folders will be created automatically
-if they do not exist yet.
+Usage of this script could be minimal, since you only need to extract the
+letters carefully and succesfully once. Then other scripts in this list can use
+the extracted images. Most likely the other scripts will use caching to speed
+up the system to. But in short, the script will create images of a single
+character based on a given dataset of license plate images and corresponding
+xml files. If the xml files give correct locations of the characters they can
+be extracted. The workhorse of this script is $plate =
+xml_to_LicensePlate(filename, save_character=1)$. Where
+\texttt{save\_character} is an optional variable. If set it will save the image
+in the LearningSet folder and pick the correct subfolder based on the character
+value. So if the XML says a character is an 'A' it will be placed in the 'A'
+folder. These folders will be created automatically if they do not exist yet.
+
 \subsection*{\texttt{load\_learning\_set.py}}
 
 

src/data.py

@@ -3,7 +3,7 @@ from cPickle import load, dump
 
 
 DATA_FOLDER = 'data/'
-IMAGES_FOLDER = '../images/LearningSet/'
+IMAGES_FOLDER = '../images/characters/'
 RESULTS_FOLDER = 'results/'
 
 

src/xml_helper_functions.py

@@ -6,6 +6,8 @@ from Character import Character
 from GrayscaleImage import GrayscaleImage
 from NormalizedCharacterImage import NormalizedCharacterImage
 from LicensePlate import LicensePlate
+from data import IMAGES_FOLDER
+
 
 # Gets the character data from a picture with a license plate
 def retrieve_data(plate, corners):
@@ -38,6 +40,7 @@ def retrieve_data(plate, corners):
 
     return data
 
+
 def get_transformation_matrix(matrix):
     # Get the vector p and the values that are in there by taking the SVD.
     # Since D is diagonal with the eigenvalues sorted from large to small
@@ -48,6 +51,7 @@ def get_transformation_matrix(matrix):
 
     return inv(array([[p[0],p[1],p[2]], [p[3],p[4],p[5]], [p[6],p[7],p[8]]]))
 
+
 def pV(image, x, y):
     #Get the value of a point (interpolated x, y) in the given image
     if not image.in_bounds(x, y):
@@ -67,6 +71,7 @@ def pV(image, x, y):
         + image[x_low , y_high] / x_y * a * d \
         + image[x_high, y_high] / x_y * c * d
 
+
 def xml_to_LicensePlate(filename, save_character=None):
     plate   = GrayscaleImage('../images/Images/%s.jpg' % filename)
     dom     = parse('../images/Infos/%s.info' % filename)
@@ -100,10 +105,10 @@ def xml_to_LicensePlate(filename, save_character=None):
             data  = retrieve_data(plate, corners)
             image = NormalizedCharacterImage(data=data)
             result.append(Character(value, corners, image, filename))
-        
+
             if save_character:
                 character_image = GrayscaleImage(data=data)
-                path       = "../images/LearningSet/%s" % value
+                path       = IMAGES_FOLDER + value
                 image_path = "%s/%d_%s.jpg" % (path, i, filename.split('/')[-1])
 
                 if not exists(path):
@@ -114,15 +119,19 @@ def xml_to_LicensePlate(filename, save_character=None):
 
     return LicensePlate(country, result)
 
+
 def get_node(node, tag):
     return by_tag(node, tag)[0].firstChild.data
 
+
 def by_tag(node, tag):
     return node.getElementsByTagName(tag)
 
+
 def get_attr(node, attr):
   return int(node.getAttribute(attr))
 
+
 def get_corners(dom):
     p = by_tag(dom, "point")
 
@@ -134,4 +143,4 @@ def get_corners(dom):
     return get_attr(p[0], "x") - x, get_attr(p[0], "y") - y,\
            get_attr(p[1], "x") + x, get_attr(p[1], "y") - y,\
            get_attr(p[2], "x") + x, get_attr(p[2], "y") + y,\
-           get_attr(p[3], "x") - x, get_attr(p[3], "y") + y
+           get_attr(p[3], "x") - x, get_attr(p[3], "y") + y