Moving object detection and classification using neural network

Dewan, M. Ali Akber, Hossain, M. Julius and Chae, Oksam (2008) Moving object detection and classification using neural network. In Agent and Multi-Agent Systems: Technologies and Applications - Second KES International Symposium, KES-AMSTA 2008, Proceedings. vol. 4953 LNAI, pp. 152-161 . (doi:10.1007/978-3-540-78582-8_16).

Abstract

Moving object detection and classification is an essential and emerging research issue in video surveillance, mobile robot navigation and intelligent home networking using distributed agents. In this paper, we present a new approach for automatic detection and classification of moving objects in a video sequence. Detection of moving edges does not require background; only three most recent consecutive frames are utilized. We employ a novel edge segment based approach along with an efficient edge-matching algorithm based on integer distance transformation, which is efficient considering both accuracy and time together. Being independent of background, the proposed method is faster and adaptive to the change of environment. Detected moving edges are utilized to classify moving object by using neural network. Experimental results, presented in this paper demonstrate the robustness of proposed method.

This record has no associated files available for download.

Contributors

Author: M. Julius Hossain

Download statistics