Combining appearance and geometric features for facial expression recognition

Abstract

This paper introduces a method for facial expression recognition combining appearance and geometric facial features. The proposed framework consistently combines multiple facial representations at both global and local levels. First, covariance descriptors are computed to represent regional features combining various feature information with a low dimensionality. Then geometric features are detected to provide a general facial movement description of the facial expression. These appearance and geometric features are combined to form a vector representation of the facial expression. The proposed method is tested on the CK+ database and shows encouraging performance.

Real-Time Facial Affective Computing on Mobile Devices

“…Previous works mainly focused on the categorical model based on six basic facial expressions (Happiness, Sadness, Fear, Anger, Surprise and Disgust) defined by Ekman et al [2]. The extracted features are applied to the classifier such as support vector machines (SVM) [12,13], AdaBoost [14] and hidden Markov models (HMMs) [15] to achieve facial expression recognition. Traditional methods mainly depend on hand-crafted features based on facial information such as geometry, appearance or texture information.…”

Novel deep learning model for facial expression recognition based on maximum boosted CNN and LSTM

“…Compared to geometric‐based methods, an appearance‐based method is robust to noise and extracts appropriate discriminative features. In few cases, hybrid‐based features are used which combines both the methods to yield better recognition accuracy [14, 15]. All aforementioned methods are suitable for images in a constrained environment.…”

Automatic Affect Perception Based on Body Gait and Posture: A Survey

“…The work conducted by Zacheratos et al [59] is one of the few studies in which global features are applied to the whole body, which, as mentioned previously, achieved one of the highest accuracies with four different emotions observed. A combination of local and global features is used in object recognition [88] and this method is also used in action recognition [89] and, more recently, in facial expression recognition with encouraging results [90]. To date, however, this approach has only had limited application to automatic affect recognition from gait and posture.…”