Multimedia Forensics - An approach to detect and analyze Human faces in multimedia files

Abstract

Manual Security is being replaced by video surveillance systems by the advent of multimedia forensics. Face detection is an important technique in multimedia forensics. Many proposed methods identify and localize the faces in an image or in an image sequence. To build fully automated systems that analyze the information contained in face images, robust and efficient face detection algorithms are required. Since, innumerable video data that is collected from video surveillance are to be automatically processed and get valuable information out of it, we need to scale the processing framework accordingly. In this paper, we propose face recognition system and minimize the overall processing time of large data set using parallel processing method. Given a video from a video surveillance, we need to convert it to frames, detect the human faces in these frames and recognize the face by comparing with a given face or comparing against database entries where criminal data is present. We use cascade classifier that takes a vector of features (decision variables) and results the probability that the vector belongs to the class. Using cascade classifier, we identify the faces from a given image and finally compare the faces with database entries.