CS 4475 Project 3

For our project, we took a collection of our group’s pictures from our travels and used feature detection of a video we took in our neighborhood at Arc de Triomf to replace the feature points with our pictures. In order for the video to look more similar to the video we found the average pixel color around the feature points and tinted our images where it was needed. This created a cohesive 3-D video mosaic of our travels whilst including a familiar landmark we passed everyday in Barcelona.

We loaded a video file from our trip, extracted its frames, and converted each to grayscale. Using a FAST feature detector, we detected and sorted keypoints in each frame. For each keypoint, we selected a small image, tinted it based on the color values from the original frame, and overlaid it on the grayscale frame. This process was repeated for all frames. We then saved the processed frames to a new video file, combining the original and processed frames side by side, resulting in a final video that visually represented the keypoints with color-tinted overlays.