Facial Expression Recognition Using Histogram Sequence of Local Gabor Gradient Code-Horizontal Diagonal and Oriented Gradient Descriptor

摘要

This paper present a original method for facial expression recognition, which fused with the Gabor filter and Local Gradient Code-Horizontal Diagonal (LGC-HD) as well as Histogram of Oriented Gradient (HOG). This approach firstly is used Viola-Jones algorithm to resize the facial expression image and convolve the facial expression image with Gabor filters to extract the Gabor Coefficients Maps (GCM). Then, we obtain Average Gabor Maps (AGM) by folding GCM of four orientations in each scale to reduce dimensions. The LGC-HD and HOG is applied on each AGM to obtain the LGGC-HD-HOG descriptor. At last, the Support Vector Machine (SVM) is adopted as classifier. We conclude that the method in this paper is better in recognition rate than other similar methods by analyzing the experimental result.