1. Here we will design our own k-means clustering algorithm. Write a program to perform the following steps:
i. Read the image 'peppers.jpg' [I=imread('peppers.jpg');]
ii. The RGB intensity component of each pixel is used as a feature vector. Now perform k-means clustering for K=3 clusters. Plot the scatter plot of the intensity values in 3D showing the 3 clusters and their centers clearly. Report the cluster centers.
iii. Show the image where pixels from each cluster belong to a single color plane (RGB). Identify which cluster corresponds to red, which one is blue and which one is green, respectively.
iv. Now perform RGB2HSV conversion of the image and use the 3D feature vector as the HSV values. Repeat k-means clustering for all the pixels. Show the scatter plot of the 3 clusters in a 3D plot and report the new cluster centers. Show the image where pixels of each cluster belong to a separate RGB plane. Comment on the difference between part iii and iv.
2. Develop an algorithm to separate the different parts of the house image below e.g. windows, roof, wall, door. Comment on your strategy and justify your reason to select this strategy.
Attachment:- Assignment Files.rar