DeitY-TU face database: its design, multiple camera capturing, characteristics, and evaluation

作者：Bhowmik Mrinal Kanti^*; Saha Kankan; Saha Priya; Bhattacharjee Debotosh

来源：Optical Engineering, 2014, 53(10): 102106.

DOI：10.1117/1.OE.53.10.102106

• face recognition
• DeitY-TU face database
• database characteristics
• soft biometrics
• anthropometry
• classification

The development of the latest face databases is providing researchers different and realistic problems that play an important role in the development of efficient algorithms for solving the difficulties during automatic recognition of human faces. This paper presents the creation of a new visual face database, named the Department of Electronics and Information Technology-Tripura University (DeitY-TU) face database. It contains face images of 524 persons belonging to different nontribes and Mongolian tribes of north-east India, with their anthropometric measurements for identification. Database images are captured within a room with controlled variations in illumination, expression, and pose along with variability in age, gender, accessories, make-up, and partial occlusion. Each image contains the combined primary challenges of face recognition, i.e., illumination, expression, and pose. This database also represents some new features: soft biometric traits such as mole, freckle, scar, etc., and facial anthropometric variations that may be helpful for researchers for biometric recognition. It also gives an equivalent study of the existing two-dimensional face image databases. The database has been tested using two baseline algorithms: linear discriminant analysis and principal component analysis, which may be used by other researchers as the control algorithm performance score.

• 出版日期2014-10