Edit me

26.1. NumPy and Scikit-Image

scikit-image is an open-source package for image processing. It has a wide range of algorithms that allow you to process your images. Documentation can be found here.

In this lesson, we are going to take a look at how to load an image into scikit-image and examine the NumPy array that results.

Installation

To install scikit-image:

python -m pip install scikit-image

Loading an Image

First, we need to get the photo we are planning to use. In this case, the photo is one I took of some flowers.

/home/cloud_user/notebooks

curl -O https://raw.githubusercontent.com/linuxacademy/content-using-pythons-maths-science-and-engineering-libraries/master/flowers.jpg/home/cloud_user/notebooks/lesson1.py```

# %%
import numpy as np
from skimage import io

img = io.imread('flowers.jpg')
io.imshow(img)
type(img)
img.shape

We should make a backup of the picture before proceeding. The file extension (.jpg, .tiff) tells Jupyter the format in which it will be saved.

Let’s make a backup of the picture:

io.imsave("img_bup.jpg", img)
io.imsave("img_bup.tiff", img)

26.2. Image Data Types

This lesson continues from our previous lesson on NumPy and Scikit-Image, and we’ll review how data is stored and data types.

In case you have shut your server down, please reload the image file as we did in the previous lesson.

# %%
import numpy as np
from skimage import io

img = io.imread('flowers.jpg')
io.imshow(img)
img.shape
img[0][0]

The image data is composed of arrays of 3 representing the RGB values.

Notice this data type (dtype) is uint8. The u means unsigned, the int means integer, and the 8 means 8-bit. In other words, the data is a number between 0 and 255. These are the values used to represent colors in the RGB (Red, Green, Blue) color model.

26.3. Transforming Images

This lesson continues from the previous lesson and reviews how we can work with the image and colorspaces.

In case you have shutdown your server, re-load the image:

# %%
import numpy as np
from skimage import io

img = io.imread('flowers.jpg')
io.imshow(img)

As you can see, the image is shown with axes values. These correspond to img[i][j], where i represents the vertical axis, and j represents the horizontal axis.

img[250][1500]

As such, we can change the color of any pixel just by accessing it and changing the three-number array. Let’s create a banner along with the top showing blue, green, and white. We can split these by x values:

for i in range(500):
    for j in range(1000):
        img[i][j][0] = 51
        img[i][j][1] = 0
        img[i][j][2] = 255
    for k in range(1000, 2000):
        img[i][k][0] = 204
        img[i][k][1] = 255
        img[i][k][2] = 51
    for l in range(2000, 3000):
        img[i][l][0] = 255
        img[i][l][1] = 255
        img[i][l][2] = 255

io.imshow(img)
img = io.imread('flowers.jpg')
io.imshow(img)

Even though this picture is a rectangle, we can put a circular mask around it using the Pythagorean theorem.

nrows, ncols, nsize = img.shape
row, col = np.ogrid[:nrows, :ncols]
cnt_row, cnt_col = nrows / 2, ncols / 2
outer_disk_mask = ((row - cnt_row)**2 + (col - cnt_col)**2 > (nrows / 2)**2)
img[outer_disk_mask] = 0

io.imshow(img)

Let’s examine colorspaces. We will be making use of the convert_colorpace method. If you review the documentation on this method, you will find a list of colorspaces that can be used.

from skimage import color

img = io.imread('flowers.jpg')
io.imshow(img)
hsv = color.convert_colorspace(img, 'RGB','HSV')
hsv.shape
hsv[0][0]
io.imshow(hsv)
Tags: python