Facial Biometric System for Recognition using Extended LGHP Algorithm on Raspberry Pi
This research provides an incremental improvement in facial recognition accuracy for biometric access control systems, particularly for applications on resource-constrained IoT devices like Raspberry Pi.
This paper proposes a facial biometric system using a novel local descriptor-based parallel algorithm on a Raspberry Pi 3 Model B. The system utilizes an extended version of the Local Gradient Hexa Pattern (LGHP) algorithm, which demonstrates improved accuracy over state-of-the-art descriptors like LDP, LTrP, MLBP, and LVP on challenging facial image databases.
In todays world, where the need for security is paramount and biometric access control systems are gaining mass acceptance due to their increased reliability, research in this area is quite relevant. Also with the advent of IOT devices and increased community support for cheap and small computers like Raspberry Pi its convenient than ever to design a complete standalone system for any purpose. This paper proposes a Facial Biometric System built on the client-server paradigm using Raspberry Pi 3 model B running a novel local descriptor based parallel algorithm. This paper also proposes an extended version of Local Gradient Hexa Pattern with improved accuracy. The proposed extended version of LGHP improved performance as shown in performance analysis. Extended LGHP shows improvement over other state-of-the-art descriptors namely LDP, LTrP, MLBP and LVP on the most challenging benchmark facial image databases, i.e. Cropped Extended Yale-B, CMU-PIE, color-FERET, LFW, and Ghallager database. Proposed system is also compared with various patents having similar system design and intent to emphasize the difference and novelty of the system proposed.