Análise de técnicas de pré-processamento de imagem para reconhecimento facial baseada em VGG Faces e Ball tree
Abstract
In recent years, facial recognition has become widely present in many devices and systems. However, despite its practicality, the performance of facial recognition is affected by factors such as lighting variation, pose, facial expression and quality of the camera that performs the capture. Therefore, it is necessary to use image pre-processing techniques to deal with these problems. Thus, in this paper, a comparative study of 6 pre-processing methods is performed (grayscale, Gaussian filter, median filter, linear filter, histogram equalization and logarithmic transformation). The tests were performed using the VGG Faces descriptor, and Ball tree as the recognition method. In addition to the bases in their original form, each base was modified by adding artificial noises, which provided new tests that allowed us to investigate the influence of each pre-processing technique in relation to the noise present in the images. The experiments suggest that the use of the median filter produces better results in images with noise such as Salt and Pepper.
References
Bah, S. M. and Ming, F. (2020). An improved face recognition algorithm and its application in attendance management system. Array, 5:100014.
de Oliveira Junior, L. L. and Thomaz, C. E. (2006). Fei face database. https://fei.edu.br/~cet/facedatabase.html. Acesso em: 09/03/2021.
Han, H., Shan, S., Chen, X., and Gao, W. (2013). A comparative study on illumination preprocessing in face recognition. Pattern Recognition, 46(6):1691–1699.
Jain, R. (1995). Machine Vision. McGraw-Hill.
Jesus, E. O. and Jr, R. C. (2015). A utilização de filtros gaussianos na análise de imagens digitais. Proceeding Series of the Brazilian Society of Computational and Applied Mathematics, 3.
Kim, J.-Y., Kim, L.-S., and Hwang, S.-H. (2001). An advanced contrast enhancement using partially overlapped sub-block histogram equalization. IEEE Transactions on Circuits and Systems for Video Technology, 11(4):475–484.
Lee, H., Park, S.-H., Yoo, J.-H., Jung, S.-H., and Huh, J.-H. (2020). Face recognition at a distance for a stand-alone access control system. Sensors, 20(3).
Luo, M., Cao, J., Ma, X., Zhang, X., and He, R. (2021). Fa-gan: Face augmentation gan for deformation-invariant face recognition. IEEE Transactions on Information Forensics and Security, 16:2341–2355.
NLPR-CBSR (2020). Casia-webface cropped. [link]. Acesso em: 09/03/2021.
Parkhi, O. M., Vedaldi, A., and Zisserman, A. (2015). Deep face recognition. British Machine Vision Association.
Savvides, M. and Kumar, B. V. (2003). Illumination normalization using logarithm transforms for face authentication. In Kittler, J. and Nixon, M. S., editors, Audio-and-Video-Based Biometric Person Authentication, pages 549–556, Berlin, Heidelberg. Springer Berlin Heidelberg.
UWA (2013). Uwa hyperspectral face database. [link]. Acesso em: 09/03/2021.
