Abstract

In ice-hockey games, the hockey stick is usually deformed significantly when the players hit the puck. Different materials are used to make hockey sticks. Studies have been proposed to find how different hockey sticks and their deformation affect the performance of players. The reconstruction of a hockey stick's deformation during a shot is the key of many of those studies. The goal of this thesis is to model the deformation of the sticks from player shot videos using physics-based models and compare two physics-based methods for their performance on this problem. In this thesis, we propose two physics-based models to deform the stick and we integrate the models into a pipeline for stick reconstruction. The contribution of this thesis is twofold: 1) Integrate the physics-based models to the pipeline including data pre-processing, denoising, and establishing constraints. 2) Adapting, implementing, and comparing two physics-based models. We evaluate the results by overlapping the deformed template with reconstructed point clouds and comparing our results with the data from a MOCAP system.