基于关键部位定位的人脸表情识别的研究与应用-计算机应用技术专业论文.docxVIP

上海交通大学硕士学位论文 上海交通大学硕士学位论文 万方数据 万方数据 基于关键部位定位的人脸表情识别的研究与应用 摘 要 移动互联网飞速发展，E-learning 借助这个平台也发展出了移动学 习的新模式。移动设备上的远程教育给学习者带来了更加便捷的学习 体验。然而由于教学者和学习者在时间上和空间上相互隔离，缺少必 要的交流，而教学者无法根据学习者的反馈尤其是学习状态来调整教 学策略，导致无法有效提高教学质量。作为肢体语言的人脸表情反映 了人的内心状态和情感状态，在人际交往中起到了重要的作用。通过 移动设备采集学习者人脸做表情识别并将分析结果反馈给学习系统或 教学者可以解决上述问题，增进师生间的情感交流。 人脸表情识别通常包括三个步骤：人脸检测及关键部位定位，表 情特征提取和表情分类。本文介绍在移动学习环境下人脸表情识别系 统的研究与应用。本文主要做了以下工作：1. 基于改进的主动形状模 型对人脸关键部位特征点进行定位；2. 分析关键部位的表情特征和人 脸表情之间的关系，提取出改进的人脸形状特征信息和人脸纹理特征 信息，并使用支持向量机分类器进行表情分类；3. 设计实验区分与人 相关和与人无关的场景，分别为两种场景寻找最佳的人脸表情特征， 相应得到 92%和 78%的识别率；4. 基于客户端-服务器架构开发一个 移动学习下的表情识别系统。. 关键词：移动学习，人脸表情识别，主动形状模型，支持向量机 I RESEARCH AND APPLICATION OF FACIAL EXPRESSION RECOGNITION BASED ON LOCATIONS OF KEY PARTS ABSTRACT M-Learning (Mobile Learning), a new E-Learning model, has become more and more popular, with the help of rapidly developed mobile technologies. It brings students tremendous convenience, but at meantime, poses new challenges. The lack of feedback has been a tough problem in E-Learning for teachers and students are separated in time and space. It is even more difficult in M-Learning, especially detecting the learning states which can be used to adapt the teaching strategy to improve the learning performance. As one form of body language, facial expression is so effective to reflect human’s inner state and emotional state that it plays a significant role in human communication. Taking photographs of the student faces from mobile devices, recognizing their facial expressions and sending them back to the learning system or teachers is potentially helpful to solve the problem mentioned above. Generally facial expression recognition involves three procedures: face detection, expression features extraction and expression classification. This thesis introduces the research and application of facial expression recognition system in M-Learning. The main effort includes: 1. Locating key facial positions based on the improved Active Shape Model. 2. Analyzing the relationship betwe