Motion estimation and object tracking in image sequences is one of the fundamental areas in image processing and continues to be researched. This book presents an approach to improving the accuracy and runtime of local optical flow methods using partial global motion models. The first part introduces concepts, assumptions, and challenges of optical flow computation, as well as several methods. The second part demonstrates various strategies for initializing optical flow computation with the help of partial global motion models. The third part evaluates these strategies in terms of accuracy and runtime using different datasets. Finally, the results are summarized and potential further investigations are suggested.