프라이버시 보호를 위한 지능형 영상 감시카메라 시스템에 관한 연구

Abstract

Recently, as the information society advances, public organizations are adopting closed-circuit television(CCTV) for the sake of administrative conveniences and social order. Likewise, the locations and quantity of CCTV installations by public organizations can be easily checked but the private installations can’t be checked since there is no restriction in the installation. Although surveillance systems such as CCTV are used to monitor illegal behaviors, they have problems such as exposing the individual privacy of innocent people. This thesis proposes the intelligent video surveillance system that can’t only prevent the privacy invasion but also complement the surveillance function reduced by privacy protection. In order to find human of privacy protection, human identification is needed in surveillance system for the prevention of intelligent and diversified crimes. This thesis proposes the human identification technique under general surveillance system environment, which certificates the identification at the time of entry. Though being available for human identification, color or height are not appropriate for extracting eigen-data for identification. For this reason, there is a limit in using these factors for the video surveillance system. In order to complement this problem, the proposed system uses smartcard system. Using the camera and the smartcard, it extracts reliable eigen-data. For the purpose of this thesis, the proposed system extracted representative colors of the clothing region for analysis using the octree-based color quantization technique. Heights were measured using the pre-entered heights of reference objects and the calculation of projective geometrical information of images such as vanishing points and lines. The proposed system extracts candidate data using the extracted color information and selects the final data from the height information of the extracted candidate information, for identification. From the experiment, we could see that the identification of a human can be checked through the proposed system and also, we have confirmed that the privacy protection function and surveillance function could be provided at the same time by using this. Also, by using the privacy protection, proposed system satisfies both functions of privacy protection and surveillance.