| Summary: | 博士 === 國立成功大學 === 電機工程學系 === 103 === The capacitive touch panel (CTP) has attracted a significant amount of interest and achieved considerable penetration of the consumer electronics products market in recent years owing to its sensitivity, excellent durability and multi-touch functionality. However, the CTP is easily affected by noise produced by the trembling of a finger, environmental magnetic interference, display noise, or process variation. Moreover, when a user draws with different speeds, the measurement noise caused by the sensor IC induces an error in the touched position and zigzagged trajectory, especially when the motion is slow. Although the well-known moving average filter (MAF) method is frequently used to reduce the measurement noise, it needs a large number of points in a specific interval to filter out a significantly high frequency noise, leading to trajectory delay and amplitude decay.
This dissertation proposes three novel touch algorithms and verifies their effectiveness by experiment and measurement results. The touch algorithm of Kalman filter (KF) firstly is adopted to reduce the noise effect, and is combined with the stroke reconstruction algorithm to detect touch pressure without increasing hardware costs or the need for a power source for the stylus. The results of experiments on the proposed CTP system were analyzed, demonstrating the effectiveness of the proposed stylus and its stroke reconstruction algorithm. Moreover, the robust tracking algorithm of the particle filter (PF) as second touch algorithm is utilized to overcome the problem of modeling error in the KF method, which accurately estimates the touched position and trajectory when the touch movement changes rapidly with a nonlinear trajectory. Experimental results demonstrate that regardless of linear and nonlinear scenarios, the PF offers better root mean square error (RMSE) of linear and nonlinear tracking trajectories than that of KF. Furthermore, in the third touch algorithm, to reduce the computation cost and maintain the trajectory smoothness, the algorithm based on KF of the mixed strategy is proposed by using the fuzzy logic-based adaptive strong tracking Kalman filter (FLASTKF), which effectively mitigates the effect of variation of measurement noise and supplies accurate estimation of the touched position. In particular, this work also provides a novel method to measure and quantify the smoothness of a touched trajectory. The experimental results indicate that the proposed FLASTKF method successfully achieves the a smooth tracking trajectory, regardless of speed, as well as decreases the mean tracking error by 85.4% over that achieved using the MAF.
|
|---|
