Character read system

Abstract

Abstract PURPOSE: To improve read precision without reference to the state of a character by setting three slice levels, by calculating the circumferential length of a character pattern and the area of a character part, and then by checking and deciding the propriety of the character. CONSTITUTION: Pattern information obtained by scanning and photoelectric conversion by photoelectric conversion part B is digitized by three slice levels and stored in memory C. Binary-coding part D selects one of slice levels and converts it into a binary signal. On the basis of this binary-coded signal, calculation part E calculates the circumferential length of a pattern and the area of a character part. Character decision part F and final decision part G select the center slice level of the character pattern and perform character recognition processing on the basis of the binary-coded information. Next, a check on whether the ratio of area to circumferential length of the decided character is within a predetermined adequate range is made; when it is within the range, the above-mentioned result is regarded as the final decision result and when not, slice levels are changed to make a decision again. Thus, even when the character is binary-coded darkly or lightly, the accurate decision can be made lastly. COPYRIGHT: (C)1981,JPO&Japio

Inventors

• 👤 Kenji Otsu

Bibliographic Details