Optimizing 3D localization of UAV using multiple camera -- 2

Couple of ideas are explained in the uploaded document. We can discuss over them to choose one before start working on it. There are two approaches: 1) The first one is to find correct correspondences between the images from the 2 cameras, so that we can align the images from the 2 cameras and perform a longer feature tracking. 2) The second approach is to fuse the odometry estimated from the 2 cameras and fuse them in a designed common reference frame and transfer back the fused result back to the 2 cameras. (This method is explained properly in the scientific paper (named KazikCVPR12), which is also uploaded for reference) Once agreed for the work the data (image sequence) will be provided, although a sample is also uploaded.