CONTOUR EXTRACTION FROM MRI IMAGES FOR HUMAN TONGUE

摘要：

In this article, we present the results on employing the Active Appearance Model to extract speech articulatory feature from Magnetic resonance imaging images automatically. The movements of tongue are recorded by MRI from Chinese subjects. The features of speech articulatory organs include both texture and shape information. An AAM based framework is proposed to deal with the high nonlinear property of tongue deformation during articulation, which demonstrates the advantage for tracking tongue shape from high noised Magnetic resonance imaging images. The performance of this framework was evaluated by comparing with manually labeled contour with automatically extracted ones. The average error is around 2 pixels.

关键词： Tongue Contour MRI images speech Production

作者: Jianguo Wei Jianwu Dang

作者单位: School of Computer Science and Technology Tianjin University Weijin Road, Tianjin, 300072, China School of Computer Science and Technology Tianjin University School of Information Science Japan Adv

会议类型: 国际会议

会议名称: 2010 International Conference on Circuit and Signal Processing(2010年电路与信号处理国际会议 ICCSP 2010)

会议地点: 上海

会议语种:英文

页码: 261-264

在线出版日期: 2010-12-25（万方平台首次上网日期，不代表论文的发表时间）