| DC 欄位 |
值 |
語言 |
| DC.contributor | 資訊工程學系 | zh_TW |
| DC.creator | 李怡臻 | zh_TW |
| DC.creator | Yi-Zhen Li | en_US |
| dc.date.accessioned | 2012-6-27T07:39:07Z | |
| dc.date.available | 2012-6-27T07:39:07Z | |
| dc.date.issued | 2012 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=995202078 | |
| dc.contributor.department | 資訊工程學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 在人機互動中，觸控螢幕裝置是我們用來接收資訊的主要媒介之一，在小尺寸的螢幕技術已經成熟並帶來極為便利的運作模式，相對的大尺寸的觸控面板將會是未來螢幕發展的趨勢，其中如何實現快速及客製化的開發出符合需求的觸控處理器是重要的議題。
本研究提出一個結合時空型態手勢與離散事件型態手勢的泛用途智慧型分類器，此一分類器使用機率神經網路作為時空型態手勢辨識核心，並以PSO演算法作為PNN分類器的學習機制；離散事件手勢辨識是在離散時間軸上辨識連續的手勢狀態轉移。結合這兩種型態手勢，我們使用Grafcet離散事件模型設計出手勢處理器指令集架構，提供手勢應用系統開發者可據以定義並快速生成各種客製化的人機互動界面手勢處理器。
最後我們以醫療床邊照顧系統與IPTV兩個應用為例，根據此手勢處理器指令集架構定義出兩個客製化的手勢辨識處理器。針對新世代消費型電子產品，此一智慧型分類器提供了一個具有高度設計彈性、優異的辨識性能與快速雛形化的手勢辨識與人機互動方法。
| zh_TW |
| dc.description.abstract | In human-computer interaction (HCI) system, the touch display device is one of the main medium to receive information. There are many small-size touch panel current, so the large-size touch panel is the trend . How to establish a fast and customized touch processor becomes important.
We proposed an intelligent classifier which integrates spatial and temporal pattern into discrete event modeling in gesture recognition. We use probabilistic neural network (PNN) as a kernel recognition algorithm, and adding particle swarm optimization (PSO) in the learning phase. As the discrete event modeling is to recognize the continuous-gesture-state transition in the discrete time-slice. To combine these two types of gestures, we use Grafcet discrete event modeling to design the gesture processor instruction set architecture. User can use it to enerate a variety of customized HCI interface gestures processor quickly.
With this gesture processor instruction set architecture, we can define customized gesture recognition processor such as medical bedside care system and IPTV. This intelligent classifier provides an excellent recognition performance, high flexibility and rapid prototype gesture recognition in human-machine interaction application .
| en_US |
| DC.subject | 光學式 | zh_TW |
| DC.subject | 觸控 | zh_TW |
| DC.subject | 手勢 | zh_TW |
| DC.subject | 自訂手勢 | zh_TW |
| DC.subject | touch | en_US |
| DC.subject | optical | en_US |
| DC.subject | gesture | en_US |
| DC.title | 光學式觸控人機介面的手勢分類器設計 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Gesture Classifier Design for Optical Touch Human-Machine Interface | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |