Analysis of Facial Expression Recognition of Visible, Thermal and Fused Imaginary in Indoor and Outdoor Environment

Abstract

This research study performed on visible images, thermal images and fused images for facial expression recognition. Linear Discriminant Analysis has implemented for feature extraction technique and support vector machine to calculate the result. This work is implemented on a newly designed database of 20 peoples’ facial expression which includes visible images, thermal images, and fused images. The extracted features of visible, thermal and fused images are utilized for classification using support vector machine. This study focuses on 5 types of facial expression. Better results are achieved on smile and anger expression. The comparative analysis of this study is done on visible, thermal and fused facial expression images. The experimental result analysis shows that fused images give better results as compared to visible images. The accuracy of smile expression is better than anger and disgust facial expression. The implementation is carried out on dataset designed in indoor and outdoor environmental setup.