Comparison of Different Image Enhancement Algorithm for Emotional Face Detection

The aim of this study is to identify different human facial expressions using image enhancement algorithms in Google COLAB. A dataset of images of different expressions like ‘happy’, ‘disgust’ using novel Blocking Effect Reduction Filter. 100 images were taken for sample preparation groups. 25 images of expression "happy" were taken as training dataset. 25 images of expression "sad" were taken for training dataset. 25 images of "sad" expression were taken as testing dataset and similarly, 25 images of expression "happy" were taken as testing dataset. From SPSS the Gpower was obtained as 80%. The proposed novel Blocking Effect Reduction Filter classifier has an accuracy of 98% with sensitivity of 82% and Histogram Equalization has an accuracy of 88% with sensitivity of 78% in facial expression images detection and classification. The 2-tailed significance of accuracy and sensitivity is 0.002 (p<0.05) and 0.840 (p>0.05) respectively. In Facial Expression Recognition, the proposed novel Support Vector Machine is slightly better than Histogram Equalization. This proves the innovative ability of the novel Blocking Effect Reduction Filter in real-time implementations.