Understanding Effects of Image Resolution for Facial Expression Analysis

Abstract

Recognizing spontaneous expressions in real environments often works on face images captured in lower resolutions or with larger variation of head poses. In this paper, we analyze the effects of face image resolution for performance of automatic facial expression analysis (AFEA) systems. Different approaches are compared for face acquisition, facial feature extraction and representation, and facial expression recognition. Our evaluations are conducted at five different resolutions of the head region on two databases: 1) Cohn-Kanade expression database; and 2) FABO database. There is no obvious difference in the performance of expression analysis when the head region resolution is higher than 72x96 pixels. The performance of AFEA is acceptable for head regions with a resolution of 36x48. For head regions with lower resolutions than 36x48, the performance decreases quickly and AFEA systems become impractical.

8 Figures and Tables

Cite this paper

@inproceedings{Tian2012UnderstandingEO, title={Understanding Effects of Image Resolution for Facial Expression Analysis}, author={YingLi Tian}, year={2012} }