Current infrared touch screen has problems in the multi-touch recognition and causes zigzag trace when drawing. In order to deal with these problems, an optimal algorithm based on target tracking was presented combined with Kalman filter and validation region algorithm. In data association, the validation region algorithm was used to match the right touch point's estimate value with the original target trajectory and delete the wrong touch point's estimate value. With the touch point motion model, Kalman filter was adopted for track smoothing and target movement prediction. Compared to the original recognition algorithm, the time of single touch point recognition was added about 3 μs by the optimal algorithm, but the smoothness in the trajectory angles was improved and about 60% of the burr amount was decreased. The experimental results show that some problems of the infrared touch screen are solved by the optimal algorithm, the effectiveness for multi-touch applications and trajectory smoothness is demonstrated, and the drawing effectiveness on the infrared touch screen is improved.