Side-View Face Recognition Using Enhanced Landmarks

The recent years have seen an increase in the growth of face recognition for computer aided identification. Majority of the processes and products developed including the traditional Viola Jones algorithm coupled with re ranking is successfully being applied on all frontal faces. Face detection remains a challenge because of variations in facade, illumination and expression. Problems arise particularly when searching for an image in the database containing side view face images. This study focuses on side-view profiles keeping in mind the current needs of forensics, census departments, police forces, and an array of government organization as well as the students studying data security. The literature survey explains various techniques currently in use to detect a front view face images. The statement of research problem identifies the challenges of detecting side view images. The proposed solution takes into consideration 15 automatically generated facial landmarks; mainly ears as an additional feature. Preprocessing is applied to the registered images to remove the background. A baseline algorithm is used and then a nearest neighbor searching scheme is applied.

Autorentext

Prof. Dr. Malik Sikandar Hayat Khiyal is currently a Professor at Faculty of Computer Science, Preston University, Islamabad. He remained Chairman Department of Computer Sciences and Software Engineering in Fatima Jinnah Women University Pakistan and IIU, Islamabad. Also worked 25 years in PAEC.