How to get pixel rgb values using matplotlib?

Question

I need to find all red pixels in an image and create my own image with just the red pixels. So far I have been experimenting with matplotlib as I am very new to it.

def find_red_pixels(map, upper_threshold=100, lower_threshold =50):
    """Finds all red pixels of the "map" image and outputs a binary image file "map-red-pixels.jpg" """
    red_map = []
    for row in map:
        for pixel in row:
            print(pixel)

I am brand new to image manipulating and have tried this, however I do not understand what the values [0.0.01] etc. mean which is outputted as pixels. Is there an easy way to do this?

Answer 1

The first thing to understand is the way the image is represented in an array when it is read in. Here I read in a 320x160 image of a rainbow:

>>> img = plt.imread('rainbow.png')
>>> img.shape
    (160, 320, 4)

This shows the dimensions of the array - note the last element of the tuple is 4 . These values represent the red pixel values, the green pixel values, the blue pixel values, and the transparency or alpha values.

We want to find pixels which have a certain red value, so we're just interested in the first of these. Let's split these out so we can just work with them:

>>> r = img[:,:,0]

This isolates the portion of the array relating to the redness of the pixel values, represented by a value between 0 and 1 . We now want to isolate the locations of the pixels which have a certain redness value. We can use np.where for this, finding the elements in our new array r which have values of over 0.5 (this value can be altered to fit your requirements of upper_threshold and lower_threshold ), and replacing these with a value of 1 , and 0 otherwise; creating a binary image.

>>> binary_img = np.where(r > 0.5, 1, 0)

Finally, we just need to save this image, using plt.imsave :

>>> plt.imsave('rainbow_red.png', binary_img , cmap='gray')

Giving the following result: