Our Recognition Technology
 A new recognition technology,
Machine-Learned Fragment Analysis (MLFA) has appeared as a component of
ExperVision's TypeReader 1.0, which was introduced in late 1991. This new
technology applies a totally new set of algorithms to the problem of
recognizing text from image. MLFA does not employ any form of sequential
database searching/matching methodology. Rather, it uses pure mathematical
modeling to evaluate image fragments for content and the reliability of
that content.
MLFA technology is based on a multi-dimensional
mathematical model. This model is extremely fast. It has the capacity to
contain font and character information for virtually any character in any
language. The MLFA model was initially developed to enable recognition of
Kanji (a symbolic language, used in Japan and China). The Kanji language
contains over 7000 characters, compared to the approximately 75 characters
found in English and European alphabets. The current implementation of
MLFA technology -- TypeReader 1.0 for Windows --supports the recognition
of virtually every commercially available font used for English.
| 
