Human Action Recognition Using Accumulated Moving Information

Abstract

This paper proposes a human action recognition algorithm which can be efficiently applied to a real-time intelligent surveillance system. This method models the background, obtains the difference image between input image and the modeled background image, extracts the silhouette of human object from input image, and recognizes human action by using coordinates of object, directions of that and accumulated moving regions of that. The human actions recognized in this study amount to a total of 8 type of actions, which include walking, raising an arm (left, right), raising a leg (left, right), sitting and crouching. The proposed method has been experimented for 8 different movements using 4 people using video input of a webcam and it has shown good results in terms of recognizing human action.

Action Recognition Framework using Saliency Detection and Random Subspace Ensemble Classifier

“…Due to the growing interest in human interaction and computation, research on action recognition and further kinds of research reflecting the intelligent interaction environment has been studied. It has been a growing research topic in computer vision due to progressively needs from a diversity of areas such as human-computer interfaces, video surveillance, sports video analysis entertainment environments and healthcare system [1]. Even though a number of evolutions have been achieved in action recognition, it still has challenges such as illumination fluctuations, camera motion, inter and intra class variations, etc.…”