基于隐性马尔可夫模型的手势识别设计和优化

HMM-based sketch recognition system:design and optimization

新型的笔式交互技术要求能够高效地识别用户手势,适应用户的手绘风格。建立了基于隐性马尔可夫的手势识别模型,在此基础上提出了在重采样阶段的中点补偿和编码阶段的方向编码优化方法。实验结果表明该识别模型能以更精简的采样点数量表示手势并给出良好的识别结果,减少了模型训练的运算量。

Newly pen-based interaction technology requires efficient user gesture recognition and user style adaption.This paper built an HMM-based sketch recognition system and used a new method called midpoint-compensation in resampling phase and an improved direction coding method for optimization.The experiment shows that the sketch recognition model can recognizes user gesture efficiently meanwhile with smaller data to represent the gesture and less computation for training the model.