In [1]:
from IPython.core.display import HTML
def css_styling():
sheet = '../css/custom.css'
styles = open(sheet, "r").read()
return HTML(styles)
css_styling()
Out[1]:
This exercise is designed to motivate the use of object oriented programming in scientific computation via a simplified case using images. Your task is to program the Python classes which reads in the data from a file, manipulates the data and plots the image.
In [2]:
import os
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import numpy as np
import sys
%matplotlib inline
You have been provided with some image files i.e. img1, .. img4 in the data directory in portable greymap (.pgm) format. Although admittedly a primitive image format, .pgm files are simple to manipulate as they contain only one pixel grey value per $x, y$ pixel in the image: the scale runs from 0 (black) to 255 (white). This represents a common task in scientific computing where you must read in some field data on a grid of points. You are provided with the code to read and reshape this data from a file!
Here's a demonstration of a greymap image that might be stored in .pgm format using a simple gradient of white to black - it is displayed here using a contour plot:
In [3]:
Nx = 72
Ny = 72
img_x = np.linspace(1, 0, Nx)
img_y = np.linspace(1, 0, Ny)
X, Y = np.meshgrid(img_x, img_y)
# Generate the gradient image - this could be stored in .pgm format!
img_z = (X+Y) * 255*0.5
print(img_z)
fig = plt.figure()
ax = fig.add_subplot(111, aspect='equal')
ax.contourf(img_x, img_y, img_z, 20, cmap=cm.Greys_r)
ax.set_xlabel('x')
ax.set_ylabel('y')
Out[3]:
Create a SquareImage class which reads (portable greymap, '.pgm') image data from a file. The class should implement the following:
The Initialiser method __init__(self, ...), which takes a string filename as an argument and
stores $Nx$ and $Ny$ as instance attributes, both equal to 72 (this isn't robust, but simplifies the exercise)
calculates and stores $x$ and $y$ as instance attrbutes. These are again the regularly spaced pixel $x$ and $y$ values from 1 to 0 (image colour values in pgm files start from top right pixel) - use linspace from 1 to 0 with $Nx$ and $Ny$ points respectively
Read and store image grey levels in filename as an instance attribute, $z$. The line for extracting this data from the files is the same as before,
np.loadtxt(filename, skiprows=4).reshape(self.Nx, self.Ny)
An add_image method which takes an image argument, and adds the z component of image to self.z
self as the first argument! Instance methods require us to be specificA plot method which takes no extra arguments, and plots the current instance attributes $z$ as a contour vs $x$ and $y$ (also instance attributes). As this exercise is not testing your matplotlib, we provide the code for the function here:
fig = plt.figure() ax = fig.add_subplot(111, aspect='equal') ax.contourf(self.x, self.y, self.z, cmap=cm.Greys_r) plt.show()
In [4]:
# Implement the SquareImage class here:
In [5]:
# The image file names
names = ['img1.pgm', 'img2.pgm', 'img3.pgm', 'img4.pgm']
files = [os.path.join('data', name) for name in names]
The following cell has left out some code where indicated - Fill them in!
In [6]:
# Instantiate the class and plot each picture.
imgs = []
for f in files:
# Produce an instance of SquareImage, passing 'f' as an argument
# image = ...
print(image)
imgs.append(image) # objects are first class instances: add to a list
# Use the 'plot' method to plot the instance
# ...
Now that we can read in image data from a file, extend your SquareImages class above so that if the filename is None (python keyword), we store the $z$ attribute as np.zeros([Nx, Ny]).
This will require an if statement, e.g.:
if filename is Nonestore z as zeroselseread and store z data
The default filename argument should be None, so that SquareImage() produces an 'empty' image.
In [ ]:
# Create an 'empty' SquareImage
combined = SquareImage()
print(combined.z)
Now use the add_image method of the empty image to add on the contents of all other image in the list of imgs
In [ ]:
# EXERCISE: Loop over the list of images and add on each image's z values to `combined':
# ...
# ...
# Plot
combined.plot()