設計與實作軌跡追蹤與預測演算法用於大型銀幕多點觸控介面;Design and Implementation of Trajectory Tracking and Estimation Algorithms for Large-Screen Multi-Touch Interface

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/82870

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/82870

Title:	設計與實作軌跡追蹤與預測演算法用於大型銀幕多點觸控介面;Design and Implementation of Trajectory Tracking and Estimation Algorithms for Large-Screen Multi-Touch Interface
Authors:	李青揚;Li, Ching-Iang
Contributors:	資訊工程學系
Keywords:	軌跡追蹤;軌跡預測;多點觸控;模糊邏輯;模糊類神經;紅外線相機;Human-computer interaction;Multi-touch;Trajectory estimation;Fuzzy neural network;CORDIC;IR camera
Date:	2020-01-20
Issue Date:	2020-06-05 17:38:37 (UTC+8)
Publisher:	國立中央大學
Abstract:	本論文探討設計與實作智慧型軌跡追蹤與預測演算法於大型銀幕多點觸控介面。該系統使用紅外線（IR）相機來捕獲IR光筆或手指於一個人造IR幕的反射光所產生的IR光點的座標，讓使用者輕鬆地將一個大型平面，投影屏幕或會議桌轉換為多點觸控界面，並且本系統導入單應性演算法來補償IR相機的梯形失真。本研究使用了基於CORDIC演算法的並行處理硬體架構來加速觸控軌跡的關聯性計算，並且軌跡的關聯性計算結果會更新每個已被跟踪的軌跡的運動狀態數據。這些運動數據會被用來幫助多點觸控系統評估每個跟踪軌蹟的滑動範圍，以增加軌跡跟踪運算的準確性。另外，運動數據還會被用來輔助自適應軌跡預測系統來平滑觸控軌跡。該自適應軌跡預測系統應用了模糊邏輯與模糊類神經控制，根據實測數據的分析，該系統能夠減輕測量噪聲產生的抖動軌跡，並且隨著雜訊的幅度自動減少濾波所造成的失真。為了改善用戶體驗，本系統實作於嵌入式平台，作為一個周邊界面，該系統能獨立完成所有觸控座標的計算，並且不消耗主機系統的運算資源。;This research proposes an Artificial Intelligence Multi-Touch Interface System (AI-MTIS) that demonstrates the smart trajectory tracking and estimation algorithms and the implementation of a large-screen multi-touch interface. The AI-MTIS is constructed by utilizing an infrared (IR) camera to capture IR blob that is produced from IR stylus or human finger in an emitted IR field. The coordinate of IR blob is converted to multi-touch input, and AI-MTIS allows user to easily overlay a multi-touch interface on large surface such as projector screen or conference table. The AI-MTIS is implemented as an embedded platform that does not consume computational resource from the host system. The special purpose (SP) processor with the parallel COordinate Rotation Digital Computer (CORDIC) architecture is implemented to compute observation-to-track association. The computation result of this SP processor is evaluated to update the motion status of each tracked trajectory. These motion data are included in multi-touch tracking decision, and it allows the sliding window of each tracked trajectory to be estimated for reducing the tracking error at low sampling rate. Additionally, motion data are reused by the adaptive trajectory estimation (TE) system to smooth touch trajectory and mitigate measurement noise. The proposed TE system utilizes fuzzy logic and fuzzy neural network to derive adaptive filtering decision. The filtering result shows the proposed TE system detects unwanted high frequency noise and adaptively decreases filtering gain to stabilize the smoothness of output trajectory. Overall, multiple autonomous functions are presented in this research to provide a user-friendly multi-touch interface.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	140	View/Open

社群 sharing

Loading...