BibTeX:

@article{IJIRSTV1I11139,
     title={Clustering based face recognition from videos},
     author={CHINJU POULOSE, Anu T P and Ambikadevi Amma T},
     journal={International Journal for Innovative Research in Science & Technology},
     volume={1},
     number={11},
     pages={387--391},
     year={},
     url={http://www.ijirst.org/articles/IJIRSTV1I11139.pdf},
     publisher={IJIRST (International Journal for Innovative Research in Science & Technology)},
}

Face recognition algorithms plays an important role in the field of personal identification. During the past several years video based face recognition has received significant attention. A video based face recognition algorithm usually computes a discriminative video signature as an ordered list of still face images from a large dictionary. In this proposed system discriminative video signature is generated from the combination of information present in each video frame. It facilitates in the identification of an individual as it embeds the information in the form of a ranked list of images under identical intra-personal settings from the dictionary. A combined approach, including clustering, re-ranking and fusion, is proposed for optimizing ranked lists across multiple video frames and fusing them into a single composite ordered list to compute the video signature. In clustering, each ranked list corresponding to a video frame is partitioned into different clusters and reliability of each cluster is calculated. Then the similarity scores of images within a cluster are adjusted based on the reliability of that cluster. During fusion, multiple ranked lists of a video are fused based on the adjusted similarity scores of images to generate a composite ranked list as the video signature. For matching two videos, their composite ranked lists obtained after clustering based re-ranking and fusion are compared. The discounted cumulative gain (DCG) measure is used to compare two ranked lists, which uses the ranking of images in the video signature as well as the usefulness of images in characterizing the individual in a video. The proposed video based face recognition algorithm efficiently computes the video signature and transforms the problem of video based face recognition into matching two ranked list.

Face Recognition, Clustering, Fusion, Similarity Score, Video signature, DCG