Subject Independent Automatic Facial Expression Recognition from Partially Occluded Still Images by Employing Appearance Based Feature Extraction Schemes
Naveen Kumar H N1, Jagadeesha S2

1Naveen Kumar H N, Department of Engineering and Communication Engineering, S D M Institute of Technology, Ujire, Karnataka, India.
2Jagadeesha S, Department of Engineering and Communication Engineering, S D M Institute of Technology, Ujire, Karnataka, India.

Manuscript received on 03 August 2019 | Revised Manuscript received on 09 August 2019 | Manuscript published on 30 August 2019 | PP: 3422-3428 | Volume-8 Issue-10, August 2019 | Retrieval Number: J96820881019/19©BEIESP | DOI: 10.35940/ijitee.J9682.0881019
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Facial expression recognition is the process of identifying human emotion through expressions. The world around us is constantly changing and in this ever-changing scenario, development of a system which performs automatic detection and recognition of facial expressions in a given scene is of paramount importance, and at the same time highly challenging. It becomes even more challenging when the aforesaid scene is partially occluded, thus limiting the facial area which could be explored. As such, expression recognition from partially occluded images is still largely an unexplored area. The proposed work is an attempt towards subject independent automatic expression recognition from partially occluded images. Salient features of the proposed work involves careful approximation of contributions made by facial regions like eye, mouth, and nose towards recognition of each basic expression; determination of a particular region in face which contributes the most discriminative and abstract feature for recognition of a particular expression; identification of facial region wherein expression recognition is independent of any occlusion happening with respect to that particular region. The proposed work to begin with segments facial regions from a static facial image; discriminative and abstract features extracted from so segmented facial regions are experimented upon to better understand the contribution made by each region in recognition of a facial expression. Various appearance features such as HOG, LBP and OGBP have been incorporated in the experimentation, and results obtained thereby infer that mouth region convey lion’s share of the information about probabilistic determinant of an expression and its intensity when compared to remaining regions. The proposed system outperforms holistic approaches in connection with facial expression recognition.
Keywords: Feature Extraction, Facial Expression Classification, Appearance Features, Oriented Gradients of Binary Pattern, 

Scope of the Article: Classification