Residential Burglary Prevention by Embedded Video Analysis System

• Main Authors: LU,CHENG-KUO, 呂政國
• Other Authors: HUANG,TSONG-LIANG
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/68759085587117334584

碩士 === 國立臺北教育大學 === 資訊科學系碩士班 === 104 === Residential burglary crime, does not only cause damage to the property of the victim, but can also affect the victims’ psychology producing symptoms such as fear, being afraid, anger and frustration etc. because a lot of the times, victims were in their homes when the burglary occurred. Residential burglary practices such as the ‘labeling’ technique involves the burglar going door to door during the daytime, using intercom doorbells to check who is present. If there was an absence, the burglar would stick a label or make a pen mark on the intercom button or mailbox. The next day, the burglar would either press the intercom again or directly break and enter inside with the intent to steal property. This is a common practice among burglars. With this point of view, this paper examines the use of embedded video analysis system with the Raspberry Pi device to capture human facial images to record activities facing the intercom doorbell during the day. The use of RBG color space technology when there is limited light source is useful as it is more sensitive to red and green lines, therefore reducing the influences on the image. To retrieve the correct color range, a procedure involving binarization and morphological image processing is created with oval template and eye feature detection to shape out facial features. In addition, the image is captured and an instant text message or email will alert homeowners and therefore reduce burglary in residential homes. For future references, this may be used for criminal investigations and assessments for policing public security.