System for segmenting character components
US4776024A · kind A · utility
Assignee
Inventors
Key dates
| Filing date | May 21, 1987 |
| Grant date | Oct 4, 1988 |
| Priority date | — |
| Expiry date | May 21, 2007 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06V30/10
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A system and method for electronically segmenting character components on character-containing documents involving first scanning a document and quantizing the image information obtained by the scanning into two levels, e.g., black and white, by 1 and 0 bits, and from the quantized information, m-bit OR groups are generated by sequentially ORing every m-th bit in a first direction of the quantized image, where m is an integer equal to or greater than two. The black (character) bits in each of the m-bit OR groups are counted and processed using the steps of: sequentially calculating sums of n consecutive count values by shifting one by one the count values obtained by the counting along a second direction perpendicular to the first direction; and then, segmenting character components by comparing the sums with a predetermined threshold value. The ORing operations can be performed conveniently by employing the OR instruction provided in a typical microprocessor and can attain substantially the same accuracy as conventional OCR systems employing an m.times.n mask.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.