A real-time awareness system for happiness expression based on the multilayer histogram of oriented gradients

Smiling faces are instinctive facial expressions for humans to convey their happiness. Therefore, the realization of this facial expression plays a vital role in human happiness awareness. This paper proposes a low computational system specialized for real-time happiness awareness in daily life. In the training phase, a novel feature called multilayer histogram of oriented gradients (MLHOGs) is proposed to simply represent both the orientation histogram and spatial information of edges with a small vector size. For the trade-off between computation cost and detection accuracy, the active shape model (ASM) is adopted to locate the discriminative facial features. Moreover, the ASM is more flexible than a conventional statistical model. In the recognition phase, linear support vector machines (SVMs) are applied to model the MLHOG features with low training and prediction cost. The experimental result shows that the proposed system can achieve an accuracy rate of 91% for smiling face detection. Besides, neither the complex features nor a computational intensive model is adopted in this work. Moreover, the system uses only the shape features of the muscles and facial features. Such low-dimensional features can highly decrease the computational cost and allow the system to detect smiling face in real time, thereby demonstrating the feasibility of the system.