Dynamic Handwriting Features Acquisition System Design

Chinese characters are based on strokes unlike western characters based on letters, so stroke feature study is the key of dynamic features of Chinese characters handwriting study. In this paper it is presented to acquire acceleration information of handwriting, to get basic stroke information of start point, end point and trace, and to create dynamic feature model of strokes. Micro-accelerometers mounted on penholder are used in this dynamic handwriting features acquisition system, and data is transmitted to computer for data processing and feature analysis through serial port after AD conversion on an MCU. In this paper, according to penholder dynamic model, principle of micro-accelerometer and rule of Chinese characters handwriting, a four-degree-of-freedom (DOF) model is introduced including three-axis accelerations and tilt angle. The tilt angle estimation is achieved by accelerometer array. Since it is difficult for a micro-accelerometer to distinct static and dynamic acceleration, the four DOF model can be used to get rid of gravity component in every measurement axis and then it is possible to calculate dynamic features of penholder. To reduce influence of mounting position error, sensor static calibration and coordinate transform are implemented. To avoid sensor noise, standard deviation of measurement data is calculated to detect start point, end point and turn point of one stroke. A dynamic feature model of strokes contains not only acceleration and tilt angle, but also velocity, displacement and stroke trace. Sensor dynamic calibrations and velocity error compensation are implemented to reduce error accumulation. Dynamic features of five typical strokes are acquired and analyzed for building the dynamic feature model of strokes. Experiment results verify that four DOF of freedom dynamic features of penholder can be obtained by using micro-accelerometer without influencing its normal motion. Standard deviation method is efficient to detect chief character points of typical strokes. It is feasible to calculate dynamic model of strokes including features of acceleration, velocity, displacement, tilt angle and stroke trace. This system can be applied in fields of calligraphy guidance or dynamic signature verification.