3D Model Of Runner Constructed From Multiple Camera Views

📄 Project Abstract

Since its inception, the aim of image processing has been to turn collections of pixels into something more powerful and recognisable; to transform the seemingly random collections of bits presented into something workable in order to extract useful data and improve our understanding of the world around us. As with all areas of computational research, increases in processing power have in turn led to more interesting possibilities and enabled us to explore our understanding of perception in its most basic form. With the release of Intel's Open Source Computer Vision Library (henceforth 'OpenCV') came a dramatic increase in the speed and efficiency with which it is possible to process images, and this project has in part been inspired by that leap forward. This project focuses on the analysis of an athlete's movement using two high-speed camera feeds, and builds a 3-dimensional model from each frame. Colour segmentation is used to find areas of skin, contour tracing to find connected re...