Research On Facial Expression Recognition Algorithm Based On Combined Features

Along with time progress and technology d evelopment, especially for the time after the battle of human versus computers between Alpha Go and Lee Sedol at Go Game, people suddenly discover that the development of artificial intelligence is out of people’s imagination. Facial expression recognition technology, which is considered as an important human-computer interaction technology, will play a vital role in the application of artificial intelligence technology. Facial expression recognition technology as the basis of human-machine emotion unstanding makes the cold machine have the ability of understanding the personal’s emotion which makes the relationship between of human and machine to get rid of the traditional Servants relationship and turn to the partnership and makes human life more warm and comfortable.In this paper, according to the character of facial expression that different region of face has a different contribution for facial expression representation; we compare and research the problem of facial expression patch selection in recent years, and then propose a novel facial expression recognition method based on combination features. The detail work of this paper is summarized as follows:1. In this paper, we firstly introduce the research background, the research significance and current research status of facial expression recognition technology.2. Combining with Facial Action Code System(FACS), we analysis the problem of the contribution of facial expression representation characterized in different facial regions and research the algorithms about facial expression patch selection and combination in the recent year.3. We introduce the face detection and facial image preprocessing algorithm about landmark detection of facial expression image, image rotation, clipping and scales normalized in detail.4. We propose a novel facial expression recognition method based on combination features. Firstly, in image preprocessing stage, we detect the landmark of facial organ and extract the Local Binary Pattern( LBP) feature of the patch around the landmark. And then Multi-Task(MTL) learning algorithm is used to learn the shared patch features which combine the discriminate features to represent the facial expression accurately across all the basic expressions. Finally, Support Vector Machine(SVM) is used to realize the classification of combination features. Experiments on the CK+, MMI, and Oulu_CASIA facial expression database show that our proposed method has a better performance.