Automated video surveillance for monitoring intrusions using intelligent middleware based on neural network

Abstract

Automated Video Surveillance Using Intelligent Middleware presented a Java based system that detects human activities in a security sensitive area and provides alarm for illegal activities identified. The system was developed using Netbeans IDE 6.8 [7] as the working environment, while Java as the programming language [8]. This study enhanced and strengthened existing security, therefore minimizing possibility of missed events which might be a threat to an area. The system composes three major processes: Motion Detection, Subject Identification and Behavior Classification. Motion Detection captures image of any movement detected. Subject Identification screens every captured image by classifying whether the motion is made by human and eliminating those which are caused by wind, animals and other non-human entity. Behavior Classification categorizes the image passed as to what action and outputs alarm if it is considered as illegal. In order to carry out these complex functionalities, a middleware was utilized to maintain continuous data flow from capturing to image processing and to reduce bulk of inputs that are being processed. Neural network [9, 10] was employed as the information processing paradigm for human or non-human and behavior classification. The result shows that the system processed video continuously as it classified behavior automatically. © 2011 Springer-Verlag.