Character recognition system

Claim 2

Character recognition apparatus for recognizing dot matrix printed characters on a document comprising: (1) means for generating an electrical binary black/white pixel image of said dot matrix printed characters to be recognized, (2) artificial fovea means for analyzing selected white pixel areas of said black/white pixel image and producing a set of measurements indicative of the boundedness of each said selected white pixel areas relative to lack pixel areas, respectively, and (3) means for utilizing said set of measurements indicative of the boundedness of said selected white pixel areas to recognize said dot matrix printed characters.

Claim 3

In a character recognition system having: image lift means for producing a black/white pixel image of characters to be recognized, means for quantizing said pixel image, artificial fovea means for emulating the behavior of a human fovea, each artificial fovea including means for temporary storage of said pixel image, means for measuring areas of said pixel image to produce image measurements corresponding to the boundedness of white pixel areas bounded by black pixels, and means for scoring said image measurements, recognition scoring means having recognition equations stored therein, said recognition scoring means being connected to said artificial fovea to score the image measurements of said artificial fovea and identify characters and produce reject signals for characters which are not identified, reject storing means for storing reject signals, learning means connected to said reject storing means and said artificial fovea, and an operator station connected to said learning means for entering to said learning means the correct identify of said not identified character and adding to the recognition equations stored in said recognition scoring means so as to avoid further rejecting of characters.

Claim 4

In a character recognition system having: image lift means for producing a black/white pixel image of characters to be recognized, means for quantizing said pixel image, artificial fovea means for emulating the behavior of a human fovea, each artificial fovea including means for temporary storage of said pixel image, means for measuring areas of said pixel image to produce image measurements corresponding to the boundedness of white pixel areas bounded by black pixels, and means for scoring said image measurements, recognition scoring means having recognition equations stored therein, said recognition scoring means being connected to said artificial fovea to score the image measurements of said artificial fovea and identify characters and produce reject signals for characters which are not identified, reject storing means for storing reject signals, learning means connected to said reject storing means and said artificial fovea, and an operator station connected to said learning means for entering to said learning means the correct identify of said not identified character and adding to the recognition equations stored in said recognition scoring means so as to avoid further rejecting of characters.