Motion History Images For Action Recognition And Understanding

Human action analysis and recognition is a relatively mature field, yet one which is often not well understood by students and researchers. The large number of possible variations in human motion and appearance, camera viewpoint, and environment, present considerable challenges. Some important and common problems remain unsolved by the computer vision community. However, many valuable approaches have been proposed over the past decade, including the motion history image (MHI) method. This method has received significant attention, as it offers greater robustness and performance than other techniques. This work presents a comprehensive review of these state-of-the-art approaches and their applications, with a particular focus on the MHI method and its variants.