Measurement of Strokes in Hand Writing Japanese Character

Abstract

Handwriting is one of the most natural ways of writing characters. A pen tablet and a mouse are generally used as computer input devices. However, the feeling of writing is extremely different between the writing with the pen tablets or the mouse and the writing on the papers with the brush pen. We propose a new computer input device. A brush pen is a traditional writing material. The path of the brush pen is inputted to computers by using a CCD camera. At first, the inputted strokes are coded into histograms of the moving direction. The inputted characters are identified by the correlation values between the histograms of the inputted characters and the histograms of the reference pattern. Additionally, the line width of the strokes written by the brush pen is approximated by the straight lines. And marginal distributions of the approximated character for the horizontal distribution and for the vertical distribution are obtained. The result of comparison about the marginal distribution is used as subsidiary parameters for the character recognition. By using the proposed system, the natural hand writing characters can be inputted into the computer. And By using the two kinds of pattern recognition method, the improvement of the pattern recognition is achieved.