权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Fusion of array signal processing and facial image process-ing for hands-free robust communication and authentication.

融合阵列信号处理和面部图像处理，实现免提稳健通信和身份验证。

基本信息

批准号：
08680443
负责人：
MATSUMURA Masafumi
金额：
$ 1.54万
依托单位：
Osaka Electro-Communication University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
1996
资助国家：
日本
起止时间：
1996 至 1998
项目状态：
已结题

项目摘要

Human speech sounds with speaker characteristics involve complex interactions of the larynx, nasal cavity, oral cavity, and oropharynx. Authentication technique can be used to verify the identity claimed by people accessing systems ; that is, it enables access control of various services by voice and face iniages.In conventional approach, however, there is few study on fusion of array signal processing and facial image processing for hands-free robust communication and authentication. The primary aim of our research is to develop new techniques for automatically recognizing who is speaking by using speaker-specific information included in speech wave and facial image. Intelligent audiovisual sensing system with four-microphone-array and CCD camera has been developed for hands-free robust communication and authentication. Array signal processing performed accurate estimation of mouth position. An articulatory model of three-dimensional vocal tract obtained by magnetic resonance imaging was proposed for extraction of speaker-specific information. A Color image processing technique was proposed for face segmentation and facial feature detection. Our group has successfully developed and tested the intelligent audiovisual sensing system for hands-free robust communication and authentication. The proposed techniques have wide application in the areas of speech communication and in the study of human interaction.

具有扬声器特征的人类语音涉及喉咙、鼻腔、口腔和口咽的复杂相互作用。认证技术可以用来验证访问系统的人的身份，即通过语音和人脸图像来实现对各种服务的访问控制，然而，在传统的方法中，很少有研究将阵列信号处理和人脸图像处理相结合来实现免提鲁棒通信和认证。我们的研究的主要目的是开发新的技术，自动识别谁是说话者通过使用特定的信息，包括在语音波和面部图像。智能视听传感系统与四个麦克风阵列和CCD摄像头已经开发的免提强大的通信和认证。阵列信号处理执行嘴位置的精确估计。提出了一种基于磁共振成像的三维声道发音模型，用于说话人特定信息的提取。提出了一种基于彩色图像处理的人脸分割和特征检测方法。本集团已成功开发及测试智能视听感应系统，用于免提稳健通讯及认证。所提出的技术在语音通信领域和人类交互的研究中具有广泛的应用。