Apparatus and method of augmenting video

摘要

A method of generating an internally consistent model of the state of a book captured in a video image comprises obtaining a plurality of pieces of evidence relating to the state of a corresponding plurality of aspects of the book in the video image, associating a quality score with each piece of evidence, generating an initial model of the state of the book wherein the state of the book is constrained by physical properties of the book and at least the highest scoring piece of evidence, and sequentially constraining the model in response to one or more successive pieces of evidence whose scores meet a respective predetermined first threshold value.