期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2017
卷号：95
期号：16
页码：3851
出版社：Journal of Theoretical and Applied
摘要：The immense growth in the area of computer vision systems made motion detection and tracking an at-tractive research topic. Video surveillance is an vital area, its applications including both indoor and out-door automated surveillance systems. In the context of smart home environments, surveillance systems have as principal end to control the safety and the security of materials and of people living in a domestic environment. The automatic analysis and understanding of behaviour and interactions is a crucial job in the design of socially intelligent video surveillance system. The automatic detection addresses several hu-man factor issues underlying the existing surveillance systems. This paper introduces a technique for mo-tion detection and tracking that incorporates several innovative mechanisms. The algorithm presented here is applicable only for binary images and it have two-step procedure. Most challenging task in any facial classification technique is the representation of face in terms of a vector. This vector provides input to a trained classifier and classifier performs final classification. Input vector should represent facial character-istics in most efficient manner such that while it contains all possible information about face. When the segmentation value becomes 1.5, could achieve 95% of tracking of the human in the real time video.
关键词：Background estimator; motion detection; image segmentation; object classification; auto threshold.
