多通道人机交互及其整合算法的研究

【作者】 张宏超；

【导师】 孙亚民；

【作者基本信息】 南京理工大学 ， 计算机应用， 2004， 硕士

【摘要】 为了改进传统的以计算机为中心的人机交互方式，多媒体技术率先应用到了计算机技术上，改善了计算机的输出界面，与之对应多通道技术的应用正在试图改变计算机的输入界面。本文在这样的背景下，对多通道技术进行了探索性的研究，在多通道整合策略设计时对任务的结构进行了改进，提出了具有复杂对象结构的任务结构，方便了程序的设计，有利于多通道整合独立于具体应用。在实现整合策略时，首先完全按照面向任务的整合思想实现了一个基本算法，其中提出了事件相关矩阵，便于程序的实现。然后针对创建任务槽时容易出错的问题，把整合策略改进成为基于任务识别的整合策略，可以充分利用用户输入的信息，并且可以利用识别过程的中间结果帮助任务槽的填充，并且收到了比较好的效果。最后对人机交互的未来和研究方向进行了展望。更多还原

【Abstract】 In order to change the traditional computer-centered mode of HCI (Human-Computer Interaction) the Multimedia technology is used in computer. And it makes the interface of output better. On the other hand Multimodal technology is changing the interface of input. In that condition Multimodal technology is studied in this thesis. The structure of the task is developed when we design the Multimodal integration strategy. The task structure with complex object structure is proposed. It makes programming easier and Multimodal integration module more independent from application. In the thesis there are two integrating strategies. The first strategy is implemented according to the thought of task-oriented and "event relation matrix" is proposed to make programming more easily. In order to avoid the shortcoming that errors often happen at creating task slot the second strategy is based on task recognition. This strategy can make full use of the uses’ information. And we can get something from the integration to help fill the task slot. That strategy leads a better result. At the end of the thesis we talk about the future of HCI.更多还原