FEATURE DESCRIPTOR FOR ROBUST FACIAL EXPRESSION RECOGNITION

摘要

Methods and systems for generating feature descriptors for robust facial expression recognition pre-process face images using Gaussian filters to smooth face images. Next, gradient-based images in M scales and N directions are generated from the pre-processed face image. Furthermore, a portion of the image corresponding to each action unit is selected from the gradient-based images. The appearance of at least one facial event is then captured within a selected portion of the image. In addition, the geometry of at least one facial event within a selected portion of the image is determined to obtain a feature descriptor for each action unit for robust facial expression recognition.