Summary: | 碩士 === 國立成功大學 === 電腦與通信工程研究所 === 100 === The automatic emotion recognition system has been a popular issue in computer vision area. With emotion recognition system, computer becomes more humanized. It also brings strong impacts on many areas such as smart living and medical area. In this thesis, I use the LBP method, which was commonly used in facial expression recognition. Furthermore, We propose a novel idea to improve the result. In traditional facial expression recognition, the researchers use Viola-Jones method to crop face from input image. However, the cropped face contains unimportant information such as hair, ear and background. Thus, ASM method was used to adjust the cropped face and keep important information. Finally, we distinguish six expressions as happiness, sadness, disgust, fear and surprise with SVM.
|