Facial Expression Recognition Using Graph Signal Processing on HOG

Abstract

The Histogram of Oriented Gradient (HOG) has been found to be an effective method for the face detection as well as the facial expression recognition. However, due to the large feature length of the facial expression, there is a challenge to decrease the size of the feature vector. A novel method is proposed using the graph signal processing (GSP) with the HOG to reduce the length of the feature vector and to increase the accuracy for recognizing the facial expression. The proposed method is demonstrated on CK+ and JAFFE database and the experimental performance is compared with the results of other methods using the same databases. The better recognition rate with the significant reduction in the feature length in the experimental results shows the effectiveness of the proposed algorithm.

Keywords:

• Facial expression
• Graph signal processing
• Histogram of oriented gradients

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Hemant Kumar Meena

Hemant Kumar Meena received the BTech in electrical engineering and MTech in information and communication technology from the Indian Institute of Technology, Delhi in 2005 under the Dual Degree programme. Presently, he is pursuing his PhD in electronics and communication engineering from Malaviya National Institute of Technology, Jaipur. His research interests include graph signal processing and image processing.

Shiv Dutt Joshi

Shiv Dutt Joshi received the BE (honours) in electrical and electronics engineering from Birla Institute of Technology, Pilani, India, in 1981, the MTech degree in communications and radar engineering, and the PhD degree, both from the Indian Institute of Technology, Delhi, India, in 1983 and 1988, respectively. Presently, he is working as a professor and Head with the electrical engineering Department, IIT Delhi. His research interests include development of fast algorithms for stochastic signal processing, speech processing, and modeling of stochastic processes. Email: [email protected]

Kamalesh Kumar Sharma

Kamalesh Kumar Sharma received BE and ME degrees in electronics and communication engineering from Malaviya National Institute of Technology, Jaipur, India, in 1990 and 2001, respectively. He completed his PhD degree from Indian Institute of Technology, Delhi in the year 2008. Presently, he is working as a professor and Head with the Department of electronics and communication engineering, Malaviya National Institute of Technology. His research interests include sampling theory of signals, signal and image processing and fractional fourier transforms. Email: [email protected]