表面聲波式觸控面板之信號處理系統

Signal processing system for SAW touch panel

柯凱仁(Kai-Jen Ke)

指導教授 ： 周元昉

國立臺灣大學/工學院/機械工程學研究所/碩士(2005年)

https://doi.org/10.6342/NTU.2005.10483

摘要

觸碰面板因為人性化介面及使用上的便捷性，應用的範圍及層面有增加的趨勢，而表面聲波式的觸控面板有優於以往電阻式、電容式、光學式觸碰面板的特點，因此本文特別針對表面聲波式觸控面板的信號處理方法加以研究改進，本文利用類神經網路來當作主要的內差方法，藉由判讀歷時曲線的取樣結果，再輔以類神經網路修正，最後建立出一套完整的位置判定方式，其特點為快速判讀結果、更佳的面板解析度、靈敏的反應等。

為了減輕系統整合於其他平台上的資源消耗，本文設計的系統利用外加的控制晶片AT89S52來作為主要的控制及運算器，藉由預先於PC上訓練出類神經網路係數，並將訓練完後的結果燒錄於控制器的內部ROM之中，最後利用控制晶片處理計算類神經網路運算。根據模擬結果，藉由類神經網路輔助位置判讀相較於未使用任何修正方法來判讀位置可以提升5倍以上的精準度，此外，對於帶有微量的雜訊的歷時信號曲線，藉由類神經網路判讀位置也能有一定的準確度。而本文實驗結果已成功實現類神經網路輔助判讀觸碰面板的觸碰位置。

關鍵字

類神經網路 ； 觸碰面板

並列摘要

For the friendly interface and easy using, the application for touch panel is increasing. SAW touch panel has more advantages than other types of touch panel, so this thesis is studied the signal process of the SAW touch panel. First of all, using neural network is the main method for estimating the receiving time signal and deciding the final position of the touch point. The system which the thesis designs is complete for determine the touch position of the SAW touch panel. The characteristics of our system are quickly determining touch position, perfect resolution, and sensitive reaction.

In order to easy off other system resources, our design uses additional control chip - AT89S52 -to process all work and operation. First of all, PC is used to train the parameters of neural network, and then the parameters are burned into internal ROM of the control chip. Finally, controller operates to find out the neural output. By the simulation results, we can promote the resolution up to five times by using neural network. And using neural network can reduce the effect of the signal noise. This system successfully constructs a complete mode for finding out precise touch position on the SAW touch panel.

並列關鍵字

Neural network ； SAW touch panel

參考文獻

參 考 文 獻

[1] http://www.3Mtouch.com

[2] Paul W. Woo, ULTRASONIC DATA CONVERTER, United States Patent Number: 3134099, May 19, 1964.

[3] Alvin M. Hlady, William C. Brown, John W. Brahan, TOUCH-SENSITIVE POSITION ENCODER, United States Patent Number: 3653031, Mar. 28, 1972.

[4] Ralf G. Johnson, David Fryberger, TOUCH ACTUABLE DATA INPUT PANEL ASSEMBLY, United States Patent Number: 3673327, June 27, 1972.