| Summary: | 碩士 === 義守大學 === 工業管理學系 === 102 === Biometric system will be one of the most important applications of the 21st century technology, biometrics-related industries can imagine an optimistic view. In recent years, with the development of social networking sites, many people will upload their own life bit by bit to the network. As the saying goes, a picture is worth a thousand words, using photographic images replace text to describe has become a habit of the modern Internet.
In order to meet users'' needs, not only the human face detection, facial expression recognition is often on adding features. If you want to express an expression, the face lower half of the mouth is an important message of judgment. Smiling face is the world''s common language, most of the people in the pictures as often photographed with a smile face. Therefore, this study aims to identify and classify the smiling face and no smiling face .
Present study is based on the Viola method to face detection and retrieve the mouth area. A small amount of geometric features combined with the Back Propagation Neural Network (BPNN) method of classification smiling face. Use angles as the geometric features of the mouth, then the features input to BPNN neural network for training and classification.
Finally, the results of training accuracy rate of 90.0% and classification recognition rate 80.0%. On the results, small amount of angle features used in this study, the samples can be effectively classify most of the faces, and has an ideal smile recognition capability.
|
|---|
