To provide an online handwriting recognizing device based on the form of a character.

This online handwriting KANJI recognizing device utilizes a conventional input method for recording the constitutional relation of character forms of KANJI as reference on the basis of decomposition into character forms. An inputted handwriting KANJI is decomposed into constitutive character forms. Afterwards, while referring to the code extraction rules of the input method, a basic character form for comparison is extracted on the basis of the character constitution method of handwriting character. Then, the code of the basic character form is compressed while using the comparison formula of the input method. The sequence code of a candidate character is provided by comparing the compressed code with the contents of input method reference. Finally, a final recognition sequence code is determined through the further comparison during post-processing. A character corresponding to the sequence code of a character font file is the recognized result and that character is outputted.