Detection and tracking of a human face from videos

Detecting and tracking an object such as face in a video is a crucial task. This book presents an analysis of the problem of initial feature point selection in the tracking context. It also points out that the standard feature point criteria are more concerned with how accurate the feature point tracking will be, rather than how robust the tracking will be. A simple method for improving the initial selection of the feature points to reduce the number of possible errors during tracking and thereby ease the demand on the algorithms that further process the positions of the tracked points is presented in this book. The method is evaluated and compared to two standard feature point selection methods on a large set of real data.