Part of our Image Processing process
sample result of constructing new scenes based on epipolar-line features
another sample result of new scene constructed
Google Street View does a good job in creating a virtual tour. However, users are constrained to limited viewing spots. Here we will intend to take a totally different approach from Google Street View in building a vision map for Johns Hopkins University Homewood campus by using epipolar features. This would be an added functionality to larger scale mappings like the works of Jeffrey Martin and his group on creating large detailed gigapixel photos. Compared to the conventional method of capturing the 360 degree view by capturing images from high towers using robotic platforms like CLAUSS industry's products, we attempted on using UAV drone to overcome the limitation on reaching high altitude to create a 2D and 3D mapping of the Homewood Campus. Also, using a UAV drone to perform aerial mapping gives us a solution to the restriction in the mobility viewers have as in London 320 Gigapan, where users can see from just one center position point. Checkout http://www.med.upenn.edu/wonlab/pano.html for a sample data.