Abstract:
There are many researches in emotion recognition of human face or body movement. Inand handwriting recognition, consequently, they seem ideal for our project in visual recognition of complex state transition of motion vector in movie scene. We use the movie scene in 4 groups of Emotion Exciting, Joy, Romance and Sadness to extract 4 features of motion vector in movie scene, Volume, Direction, Speed and Energy. We conclude that Direction feature of movie scene is not appropriate for our method. The combination of data from 3 features of motion vector, Volume, Speed and Energy is the information that provides the appropriate data of movie scene for emotion recognition by using Hidden Markov Model. And we conclude that the movie scene in group of exciting emotion is the best group to be recognized by the information of movement. In our experimental we achieve 88 percent accuracy in recognition of exciting scene by the information of 3 combined features Volume, Speed and Energy. ฺBy our method we cannot separate the movie scene in group of Romance and Joy emotion by using the information of motion vector.