This document summarizes research on image segmentation using the k-means clustering algorithm with different window sizes and image sizes. It finds that as window size increases, segmentation smoothness increases and sharpness decreases. Computational time and noise tolerance also increase with larger window sizes. Testing on images of different sizes, it finds 3x3 and 4x4 windows perform better than 2x2 for larger images. The proposed method provides improved segmentation results over existing techniques by using moments and k-means clustering with window-based feature extraction.