A Combined Approach for Emotion Recognition using Bezier Curve and Facial Expression Analysis

Faiyaz Mohammad Saif, Jabin Rubayat, Fahmida Sharmin Pranto and Md. Hosne Al Walid

Extracting and comprehension of emotion is of high significance for the collaboration among human and machine communication frameworks. The most expressive approach to show the human's emotion is through facial expression analysis. Facial Expression is an acknowledged, non-nosy additionally skillful technique for communication that has been well thoroughly considered as a plausible cooperation of such interface. This paper introduces and actualizes an extraction and acknowledgment technique for facial expression and emotion from still picture. The mean of this examination is making a Facial Expression Recognition (FER) conspire by utilizing Bezier curve and afterward checking the precision by utilizing diverse classifiers (Naïve Bayes, SVM and ANN). After taking a picture from dataset and sequentially applying skin color segmentation, largest connected component and binary image conversion, eyes and lips are isolated from the face. Subsequently Bezier curve for eyes and lips are recognized and are contrasted with those pictures that are stored in the database. Next it finds the nearest Bezier curve from the database. In this way by utilizing these technique emotions (Smile, Normal, Sad & Surprise) communicated by the human face can be distinguished effortlessly. Hence, a Bezier curve based solution together with image processing is used in classifying the emotions by using WEKA. Finally, ANN, Naïve Bayes and SVM classifiers are used to determine the accuracy of the system by training and testing the dataset.