PURPOSE: To reduce the capacity of a ROM mounted on a product without lowering the output quality irrelevantly to the number of fonts by providing basic skeleton data and font dependent data.

CONSTITUTION: For example, a Chinese character shown in the figure consists of image elements STR1-STR11. The basic skeleton data includes information on the start part, intermediate parts, end part, priority, and contiguity information of the skeleton of the image elements. The font dependent data consists of face information, image element data, and weight information. The shape of the character face of the basic skeleton data is converted by using the face information of the font dependent data. Then data on one stroke is read out of the basic skeleton data according to the priority entered in the font attached data. Then the start part contiguity information of the basic skeleton data is referenced to read data on the corresponding start part out of the font attached data and the thickness of the stroke is adjusted with the weight information. Similarly, the data on the intermediate parts and end part are accessed. Thus, data on the respective strokes are generated.