Ice hockey shot motion extraction from stereo videos

点击翻译按钮获取中文摘要

Selecting the best hockey stick for a given hockey player is an important task that is currently achieved by acquiring the stick an puck motion using expensive and difficult to operate motion capture equipment. Given the complexity and price of the task, this type of analysis is only available to professional players and only in specially designed studios. The goal of this project is to make this process ubiquitous and available to all. The goal of this project is to design a low cost system that only uses a stereo camera setup for capturing the stick and puck motion in a hockey shot, and to extract an abstracted 3D representation of the stick movement, deformation, and puck trajectory. This system is compact and mobile and therefore can be brought to hockey arenas or can be setup in retail shops where will help mainstream hockey players children and adults alike select the best hockey stick for their style. We achieve this by developing several state of the art tracking algorithms for both stick and puck. In the first phase of this project, carried out during the Engage grant, we have made significant progress towards the goal and demonstrated a research prototype and compared it agains the MOCAP system. We are collaborating closely on this project with CCM, a leading company in hockey equipment manufacturing that has a large presence in Montreal. Our proposed system will be deployed in at least 25 stores instantly giving access to a critical mass of people to our system. This has the potential of revolutionizing the stick selection mechanism, thus benefiting the company as well as the Canadian gaming industry as a whole.