An Enhanced Background Subtraction Algorithm for Smart Surveillance System Using Adaptive Gaussian Mixture Technique

Abstract

Surveillance is an important component of secure infrastructure. People need to ensure that their properties, homes, friends, families, and valuables are protected from intruders. In this article, we developed an enhanced background subtraction algorithm for smart surveillance of homes under changes in illumination. An adaptive Gaussian Mixture technique was applied to improve the conventional background subtraction technique for smart surveillance system by reducing false positives and improving the system performance in motion detection. The developed smart surveillance system has the ability to send notifications containing intruder’s image to registered email account and short message system to registered phone number whenever motion is detected. The performance evaluation of developed algorithm showed a sensitivity of 80%, specificity of 92% and an accuracy of 95.56%. This system works in conditions of illumination changes and can be adopted in motion detection for home surveillance.