So I sat down with Mark to talk about my OSS on Friday. After telling him what I had been upto, he suggested that I take my OSS in a direction which interests me rather than focusing so much on SCOPE. He suggested making a panorama out of video or possibly trying image stabilization.
I had never thought of making panoramas from video and so I started searching online to see what I could come across (see my delicious oss tag list for more information). It turns out that HP Research had a project that intelligently sampled family videos for keyframes and automatically recreated snapshots based on the videographer’s intentions. Now that is a fundamentally new, cool and useful adaption of vision technology. Unfortunately they didn’t talk too much about the algorithms used as they want to capitalize on their research. I think it would be cool to get a simplified version of the panorama code working by the end of the semester.
In order to get there, I am going to first write a program which can determine the direction and speed that the camera is panning and then work on the stitching.
Additionally, it’s interesting to see that the lead CDT developer found one of my earlier OSS posts and enjoyed my props for CDT. It’s cool that the Internet can connect people that probably would have otherwise not communicated with one and another.