This document describes a system for stitching images into a panoramic view using optical flow. It involves: 1. Feature detection using Shi-Tomasi corner detection and optical flow matching between frames. 2. Key frame selection by choosing frames that overlap well and tracking features - a new key frame is selected when less than half of features can be tracked. 3. Stitching images by computing camera pose using homography between key frames, projecting image points to a cylinder coordinate system, and texture mapping the stitched panorama onto a cylinder for viewing. The system is evaluated on sample video streams and results are shown for planar and cylindrical projections of stitched images using Shi-Tomasi