基于HMM的触点轨迹识别

针对触点轨迹运动复杂,其识别准确率较低的问题。引入隐马尔科夫模型,采用基于模型描述的方法获得轨迹的动态特性,并对各个轨迹模式建立相应的隐马尔科夫模型。首先利用训练样本得到可靠地参数,然后计算测试样本对于各个模型的最大似然概率,最后选取最大概率值对应的轨迹模式作为识别的结果。实验结果表明,用HMM模型识别触点轨迹,其识别速度较快,并对不同复杂程度的触点运动轨迹,其最高识别率可达到80%。

Trajectory Recognition of Touch Blobs Based on HMM

Aiming at complex degree of the touch blobs trajectory,and the problem of low recognition rate.Hidden Markov model is introduced in this paper,using model-based method described to obtain the dynamic characteristics of trajectories,and establishing the corresponding HMM of each trajectory models.Firstly,the training samples are used to get the credible parameters of the mode.Then,the maximum likelihood probabilities of test samples are computed to the entire trained model.Last,the maximum value is saved and the corresponding model is the recognition result.Experimental result show that use HMM to identify blob trajectory,the recognition speed is fast,and the highest recognition rate can reach 80% to different complexity of blob trajectory.