A Novel Linear Collaborative Discriminant Regression Classification and L1 Norm Based Algorithm for On-Chip Realization of Uncontrolled Face Recognition

Abstract

This paper proposes the novel algorithm that suits the on-chip realization of uncontrolled face recognition system. Face recognition has its own advantage and occupied all possible venues. It plays a key role in the identification or authentication of a person in wide range applications from unlocking the mobile phone to forensics. The biggest hindering factor is its speed when it is implemented along with other systems. The systems should be implemented as standalone devices. Design of Face recognition algorithms that fit for on-chip realization is far more important. With the existence of sophisticated System on Chip hardware and software platforms IP based system design is very much possible. In this work, Collaborative discriminates classification based face recognition is designed and implemented on Xilinx Zynq SoC platform. This proposed algorithm can be converted as an IP and the same can be reused for any design using VIVADO IP integrator. The proposed algorithm and architecture shows the improvements in speed and reduction of power consumption.