Face Recognition – Extensive Survey and Recommendations

Nowadays, the digital environment is a fast-growing and potential realm of the world. Human verification and identification can be done online. Face recognition is the competitive method and best biometric modality for human identification and recognition in comparison to voice, iris, thumb, ear, hand, and retina scans. This is a potential emerging area that required sophisticated research in both academics and industry to think of a few powerful face detection strategies making it quite possible in computer vision. Also, it’s a very challenging research area because of unconstrained environments. Though most of the existing research has provided promising solutions, some of the algorithms find it difficult to yield results under different unconstrained conditions such as lighting, expression, illuminate, pose variation, low resolution, and occlusion. This paper provides a detailed review of the past as well as current research techniques and highlights the drawbacks. Especially the model, pattern, manual, and automated feature extraction techniques have been reviewed extensively and their drawbacks are highlighted. Additionally, the performances of face recognition on the standard datasets are analyzed. Finally, recommendations are provided to overcome the existing problem faced during the time of face recognition, which will help to improve the research in the future.