A Block-Wise Deformation-Based Approach for Facial Expression Recognition

This paper presents the recognition results about exploring the combination of an approach based on non-rigid deformation with a block-wise method to describe the facial deformation caused by the muscle contractions and expansions that are presented in facial expressions. The proposed feature extraction method takes advantage of the knowledge related to which parts of the face reflects the highest deformations, so we selected 4 specific facial regions at which the appearance descriptor were applied. The most common approaches used for feature extraction are the holistic and the local strategies. In this work we present the results of using a local appearance approach estimating the correlation coefficient of the 4 corresponding landmark-localized facial templates of the expression face related to the neutral face. The results let us prove how the proposed estimation of non-rigid deformation scheme with the dominant block based on computational correlation can simply and intuitively measure the deformation parameters for some of the most relevant facial regions and how these parameters together can be used to recognize facial expressions automatically. Using a Levenberg-Marquardt Back Propagation neural network, it was obtained as the highest successful classification accuracy 93.17%.