Facial Expression Recognition via Locally Linear Embedding

Locally linear embedding is an unsupervised procedure for mapping high-dimensional data nonlinearly to a lower-dimensional space. In. this paper, we propose the facial expression recognition using LLE algorithm based on Gabor filters. In our framework, data are convolved with Gabor filters firstly, and then construct more discriminative information; finally, the neighbor of data is used to construct the embedding for classification problems. For demonstrating the effectiveness of our method, we use comprehensive comparisons and extensive experiments on facial expression recognition in comparing our method with traditional linear method PCA. Experimental results show that our method is better performance than linear method.