摘要: Abstract Gestures are hand-drawn strokes that do things. These things happen at distinctive places on the stroke. We built a gesture input filter and recognizer. The is fast, because it does few computations per point, can omit pre-filter data smoothing, wild points caused by hardware glitches removed output of filter, not many points. recognizer novel combination two traditional techniques; angle filtering multiscale recognition. Because an produce well-behaved scaled output, multi-scale treatment had to be unusual.