Analyzing the individual strokes works flawlessly with Chinese and Japanese, where the stroke order is fixed (occasionally with a few variants). If you have the stroke information and the user writes correctly you can recognize characters that even humans would fail to read from the finished glyphs.