A study on facial components detection method for face-based emotion recognition

In this paper, we propose a facial components detection method for face-based emotion recognition. First, the face region is detected through a combination of the YCbCr color model and the Maximum Morphological Gradient Combination (MMGC) image. The searching region for the facial component detection is limited to the detected face region. The facial components are detected using the histogram method, the blob labeling method, and the MMGC image. Also, a tagging operation for reliable evaluation was performed. The detection performance was evaluated using a tagging coordinate. In the experiment result, the facial components detection showed 81.4% accuracy and a 53.6 ms detection time. We show the reliability of the proposed evaluation method through a number of experiments using the tagging method.