关键词: 文字书写教学, 噪声处理, 可选步长线性插值, 点序差, 虚拟平滑

Abstract: The processing of the handwriting noise in character-writing teaching system must be classified according to whether the noise can reflect users’ writing effect. In this paper, non-standard writing state is divided into three categories, which are writing too fast （or floating writing）, writing too slowly（or pause in writing）, and illegal handwriting direction, so the corresponding noise structure can be grouped into white, black and jitter. To eliminate those noises, it uses the strategies of step optional linear interpolation, point sequence difference, virtual smooth and so on respectively. Experiment and comparison analysis show that, those noise processing strategies realize the working mechanism of treatment classification according to whether the guidances should be given when a noise appeared, thus provide a true, valid, reliable data environment for the generation of the objective, accurate guidances based on the users’ writing effect in the system.

Key words: character-writing teaching, noise processing, step optional linear interpolation, point sequence difference, virtual smooth