Open AccessMultimodal verge for scale and pose variant real time face tracking and recognition
Author(s) -
Ramkumar Govindaraj,
Э. Логашанмугам
Publication year - 2019
Publication title -
indonesian journal of electrical engineering and computer science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.241
H-Index - 17
eISSN - 2502-4760
pISSN - 2502-4752
DOI - 10.11591/ijeecs.v13.i2.pp665-670
Subject(s) - computer science , artificial intelligence , facial motion capture , computer vision , face (sociological concept) , tracking (education) , facial recognition system , field (mathematics) , outlier , frame (networking) , representation (politics) , scale (ratio) , video tracking , rank (graph theory) , face detection , pattern recognition (psychology) , video processing , mathematics , geography , social science , pedagogy , law , sociology , psychology , telecommunications , political science , cartography , combinatorics , politics , pure mathematics
In recent times face tracking and face recognition have turned out to be increasingly dynamic research field in image processing. This work proposed the framework DEtecting Contiguous Outliers in the LOw-rank Representation for face tracking, in this algorithm the background is assessed by a low-rank network and foreground articles can be distinguished as anomalies. This is suitable for non-rigid foreground motion and moving camera. The face of a foreground person is caught from the frame and then it is contrasted and the speculated pictures stored in the dataset. Here we used Viola-Jones algorithm for face recognition. This approach outperforms the traditional algorithms on multimodal video methodologies and it works adequately on extensive variety of security and surveillance purposes. Results on the continuous demonstrate that the proposed calculation can correctly obtain facial features points. The algorithm is relegate on the continuous camera input and under ongoing ecological conditions.