Article ID Journal Published Year Pages File Type
484935 Procedia Computer Science 2015 8 Pages PDF
Abstract

We propose a biometric security system to identify miscreants (Watchlist person) whose information is already provided in the database. With huge volumes of information being brought forth from these surveillance cameras, methods are needed to automatically analyze these surveillance videos. In our approach we combined Principal Component Analysis (PCA) with Nearest Neighbor replacing threshold value by confidence value. The proposed approach works with only one image per person. We are able to reduce recognition time and an alarm needs to be generated as soon as a Watchlist person is caught in the video.

Related Topics
Physical Sciences and Engineering Computer Science Computer Science (General)