We will use self-written KMeans code for a useful real-world problem. In many RGB encodings each pixel is represented by 24 bits, 8 bits for each one of the main colors (Red, Green, Blue) ranging from 0 to 255, and therefore each pixel can have more than 16 million colors. In this part, We will use K-means algorithm to reduce the number of colors to 16 (or 256) so that the color of each pixel can be represented by only 4 bits (or 8 bits). In fact, we only need to store the RGB values of the 16 (or 256) selected colors in an array, and for each pixel in the image we now need to only store the index of the color in the array. Since reducing the number of the colors result in lower quality for the image, we use K-means to find the 16 (or 256) colors that best group pixels in the image.
The original image has 800x800 pixels and with 24 bits per pixel the image requires at least 800x800x24 bits to be stored. By reducing the number of colors to 16 and 256 the image only requires about 800x800x4 and 800x800x8 bits respectively to be stored. By this point hopefully you have learnt the intuition behind the clustering and grasped a hands- on knowledge on how to use clustering for real world challenges.