Method and apparatus for separating overlapping patterns

Claim 1

A method of determining the outline of a character in a character recognition system, said character being included in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; storing the location in said segment of said first pixel; scanning those pixels in said segment within a first predetermined distance from said first pixel to detect a second pixel of said second value; storing the location in said segment of said second detected pixel; and continuing said scanning to successively detect pixels of said second value within said predetermined distance of the most recently detected pixel of said second value, wherein said scanning begins at the pixel adjacent to and clockwise relative to said most recently detected pixel of said second value from the last pixel of said first value detected immediately prior to detection of said most recently detected pixel of said second value.

Claim 2

A method of determining the outline of a character in a character recognition system, said character being included in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; storing the location in said segment of said first pixel; scanning those pixels in said segment within a first predetermined distance from said first pixel to detect a second pixel of said second value; storing the location in said segment of said second detected pixel; and continuing said scanning to successively detect pixels of said second value within said predetermined distance of the most recently detected pixel of said second value, wherein said scanning begins at the pixel adjacent to and clockwise relative to said most recently detected pixel of said second value from the last pixel of said first value detected immediately prior to detection of said most recently detected pixel of said second value.

Claim 4

A method of determining the outline of a character in a character recognition system, said character being included in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; storing the location in said segment of said first pixel; scanning the eight pixels in said segment which immediately surround said first pixel to detect a second pixel of said second value; continuing said scanning to successively detect pixels of said second value in those pixels immediately surrounding the last detected pixel of said second value; continuing said storing of the locations in said segment of said successively detected pixels of said second value until said scanning again detects said first pixel, wherein said scanning begins at the pixel adjacent to and clockwise relative to said most recently detected pixel to said second value from the last pixel of said first value detected immediately prior to detection of said most recently detected pixel of said second value.

Claim 5

A method of determining the outline of a character in a character recognition system, said character being included in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; storing the location in said segment of said first pixel; scanning the eight pixels in said segment which immediately surround said first pixel to detect a second pixel of said second value; continuing said scanning to successively detect pixels of said second value in those pixels immediately surrounding the last detected pixel of said second value; continuing said storing of the locations in said segment of said successively detected pixels of said second value until said scanning again detects said first pixel, wherein said scanning begins at the pixel adjacent to and clockwise relative to said most recently detected pixel to said second value from the last pixel of said first value detected immediately prior to detection of said most recently detected pixel of said second value.

Claim 7

A method of eliminating unwanted data in a character recognition system, said unwanted data being included with, and not joined with, one or more characters in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; storing the location in said segment of said first pixel; scanning those pixels in said segment within a first predetermined distance from said first pixel to detect a second pixel of said second value; storing the location in said segment of said second pixel; continuing said scanning to successively detect pixels of said second value within said predetermined distance of the most recently detected pixel; and rejecting pixel patterns having predetermined configurations which are not attached to said one or more characters.

Claim 8

A method of eliminating unwanted data in a character recognition system, said unwanted data being included with, and not joined with, one or more characters in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; storing the location in said segment of said first pixel; scanning those pixels in said segment within a first predetermined distance from said first pixel to detect a second pixel of said second value; storing the location in said segment of said second pixel; continuing said scanning to successively detect pixels of said second value within said predetermined distance of the most recently detected pixel; and rejecting pixel patterns having predetermined configurations which are not attached to said one or more characters.

Claim 13

A method of eliminating unwanted data from a segment of pixel information containing one or more characters and said unwanted data, said pixels being either a first value, or, a second value said method comprising the steps of obtaining the pixel pattern of said character and said unwanted data in said segment, scanning said pixel pattern to detect changes therein between pixels of said first value to pixels of said second value, generating a change bit map showing said detected changes as boundaries of said change map, generating a mask bit map from said change map, said mask bit map containing bits of a first value within said boundaries and bits of a second value outside of said boundaries, and combining said mask bit map with said pixel pattern of said character and said unwanted data to remove the pixels of said second value in said pattern which represent said unwanted data.

Claim 14

A method of eliminating unwanted data from a segment of pixel information containing one or more characters and said unwanted data, said pixels being either a first value, or, a second value said method comprising the steps of obtaining the pixel pattern of said character and said unwanted data in said segment, scanning said pixel pattern to detect changes therein between pixels of said first value to pixels of said second value, generating a change bit map showing said detected changes as boundaries of said change map, generating a mask bit map from said change map, said mask bit map containing bits of a first value within said boundaries and bits of a second value outside of said boundaries, and combining said mask bit map with said pixel pattern of said character and said unwanted data to remove the pixels of said second value in said pattern which represent said unwanted data.

Claim 16

A method of determining the outline of a character in a character recognition system, said character being included in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: (a) searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; (b) storing the location in said segment of said first pixel; (c) selecting a predetermined distance; (d) scanning those pixels in said segment within said predetermined distance from said first pixel to detect a second pixel of said second value; (e) storing the location in said segment of said second detected pixel; (f) continuing said scanning to successively detect pixels of said second value within said predetermined distance of the most recently detected pixel of said second value; and (g) if the character cannot be identified based on the outline of the character so determined, selecting a predetermined distance less than the previously used predetermined distance and repeating steps (d) through (f).

Claim 17

A method of determining the outline of a character in a character recognition system, said character being included in a segment comprising a plurality of pixels, said pixels being either a first value or a second value, said method comprising the steps of: (a) searching said segment of pixels to detect the location therein of the first pixel of said second value nearest one edge of said segment; (b) storing the location in said segment of said first pixel; (c) selecting a predetermined distance; (d) scanning those pixels in said segment within said predetermined distance from said first pixel to detect a second pixel of said second value; (e) storing the location in said segment of said second detected pixel; (f) continuing said scanning to successively detect pixels of said second value within said predetermined distance of the most recently detected pixel of said second value; and (g) if the character cannot be identified based on the outline of the character so determined, selecting a predetermined distance less than the previously used predetermined distance and repeating steps (d) through (f).