Abstract: In order to design a new practical industrial human-machine interface, a touch input system design way based on a five-point touch calibration and a median average filter algorithm is proposed. The system uses a five-point touch calibration to calculate the correction factor based on a sampling of five points of the touch screen as a touch-screen calibration parameters; based on mXT224 controller, it utilizes the average median threshold way to filter white Gaussian noise and spike noise, and then effectively corrects calculation of multi-point input, step by step uploads input coordinate information, and ultimately passes information to the upper application. The tests show that the noisy environment based on the optimization of the Android multi-touch input system can maintain a high signal to noise ratio（over 80∶1）, and support with gloves or a stylus to touch input with the response time of less than 10 ms, which is accorded with the industrial Human-Machine Interface（HMI） design requirements.

Key words: android multi-touch screen, calibration, Signal to Noise Ratio（SNR）, Human-Machine Interface（HMI）

摘要： 为了设计实用新型工业人机界面，提出了一种基于五点触摸校正和中位数均值滤波的Android多点触摸输入系统设计方法。采用五点触摸校正，根据触摸屏上采样的五个点，计算出校正系数，作为触屏校正参数；在mXT224控制器处理基础上，采用中位数均值阀限滤波，滤除触摸系统高斯噪声与尖峰噪声，并结合触摸校正参数校正输入坐标，逐级上传输入坐标信息，最终传递给上层应用程序。测试表明，基于优化的Android多点触摸输入系统在噪声很大环境下能够维持高信噪比（超过80∶1），并支持带手套或触笔触摸输入，反应时间小于10 ms，符合工业人机界面（HMI）设计要求。

关键词: android多点触摸屏, 校正, 信噪比（SNR）, 人机界面（HMI）