Automatic Extraction of Facial Regions using CLM for the Application of Face Recognition
M. Annalakshmi1, S. Md. Mansoor Roomi2, V.Karthik3

1M. Annalakshmi*, Currently Working, Associate Professor in Electronics and Communication Engineering, Sethu Institute of Technology, Kariapatti, (T.N), India.
2S. Md. Mansoor Roomi, currently working as Associate Professor in Electronics and Communication Engineering, Thiagarajar College of Engineering, Madurai, (T.N), India.
3V. Karthik,  Currently Working, Assistand Professor in Electronics and Communication Engineering, Sethu Institute of Technology, Kariapatti, (T.N), India.
Manuscript received on June 16, 2020. | Revised Manuscript received on June 27, 2020. | Manuscript published on July 10, 2020. | PP: 264-268 | Volume-9 Issue-9, July 2020 | Retrieval Number: 100.1/ijitee.I7087079920 | DOI: 10.35940/ijitee.I7087.079920
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Facial Image processing is an active research area which involves various applications such as face detection, face recognition, person identification and also demographic information collection that is age, gender and race from the face. In general, all these applications fall under either holistic approach or local approach. In holistic approach, the whole face image is used for further processing. In local approach, the face is divided into various blocks which may contains the facial regions like eyes, eye brows, mouth, nose, cheek and chin regions. In order to develop a robust algorithm against scale, illumination, pose and expression issues, a component based or part based approaches were developed. For component based approach, the extraction of various face parts needs an automatic method to crop the facial regions closest to the manual cropping. The method of automatically extracting various facial parts is addressed in this paper. The Constrained Local Model (CLM) approach is used to identify the facial landmarks which in turn used to segregate the different facial parts. The performance of proposed approach is evaluated by ground truth of the respective facial components. The correspondence between the automatic extracted facial regions and ground truth is evaluated by SIFT descriptor. Experimental results on matching manually cropped facial regions against automatically extracted regions show that the CLM approach achieves promising performance. The extraction of various facial region will be very useful in Face recognition. 
Keywords: Constrained Local Model, facial land mark localization, Face region extraction, SIFT descriptor.
Scope of the Article: RFID Network and Applications