Semantic recognition of sketch is a key research field in Pen-based Human-Computer interaction. In this paper, we propose improved algorithms of sketch recognition: for a stroke with simple semanteme, we construct a quantization array for the angle of stroke as a pattern to find a match in a basic gesture vocabulary; for a stroke with complex semanteme, we segment the complex stroke into basic stokes by the feature points, then map them to a key character string and match it with the template string in the extended gesture vocabulary on a regular expression grammar.
展开▼