Method and apparatus for automatic language determination of Asian language documents

Claim 2

An automatic language determining system for determining a language of an Asian script text portion of an image of a document, the Asian script text portion including alphanumeric elements, phonetic elements, ideographic elements or pictographic elements, the system comprising: input means for inputting a digital data signal representative of the script text portion, the digital data signal comprising a plurality of signal portions, each signal portion corresponding to one of a plurality of pixels; connected component generating means for generating at least one connected component from the plurality of signal portions; feature determining means for locating script features corresponding to at least one preselected type of script feature in at least one connected component; and language determining means for comparing a distribution of the located script features with at least one predetermined script feature distribution of a plurality of Asian languages and for determining the Asian language of the script text portion in response to the comparison.

Claim 3

The automatic language determining system of claim 1, wherein the feature determining means comprises: bounding box generating means for generating, for each at least one connected component, a bounding box of the plurality of signal portions for each said connected component; and character cell generating means for generating a plurality of character cells, each character cell comprising at least one connected component.

Claim 4

The automatic language determining system of claim 2, wherein the feature determining means further comprises optical density determining means for determining optical densities of the plurality of character cells.

Claim 5

The automatic language determining system of claim 3, wherein the optical density corresponding to each character cell of the plurality of character cells is normalized to an area of said character cell.

Claim 6

The automatic language determining system of claim 3, wherein the language determining means comprises means for generating a histogram corresponding to the optical densities of the plurality of character cells.

Claim 7

The automatic language determining system of claim 5, wherein the language determining means further comprises LDA generating means for converting, by linear discriminant analysis, the histogram to a point in a variable-space, comparing means for comparing the point with a plurality of predetermined regimes in the variable space, and output means for outputting a determined language for the script text portion in response to the comparison.

Claim 8

The automatic language determining system of claim 1, wherein the plurality of Asian languages comprises at least one of Korean, Chinese and Japanese languages.

Claim 9

An automatic language determining system for determining an Asian language of an Asian script text portion of an image on a document, the Asian script text portion including alphanumeric elements, phonetic elements, ideographic elements or pictographic elements, the system comprising: a scanner for scanning the document and outputting a digital data signal representative of the image on the document, the digital data signal comprising a plurality of signal portions, each signal portion comprising data representative of a corresponding one of a plurality of pixels of the image; a memory for storing the digital data signal; and a controller for determining the Asian language of the script text portion of the image, comprising: connected component generating means for determining at least one connected component from the plurality of signal portions; feature determining means for locating script features, corresponding to at least one predetermined type of script feature, for each at least one connected component; feature distribution determining means for determining a distribution of the located script features; comparison means for comparing the determined feature distribution to predetermined feature distributions of at least one Asian language; and output means for outputting a signal indicative of a determined language of the script text portion.

Claim 10

An automatic language determining system for determining an Asian language of an Asian script text portion of an image on a document, the Asian script text portion including alphanumeric elements, phonetic elements, ideographic elements or pictographic elements, the system comprising: a scanner for scanning the document and outputting a digital data signal representative of the image on the document, the digital data signal comprising a plurality of signal portions, each signal portion comprising data representative of a corresponding one of a plurality of pixels of the image; a memory for storing the digital data signal; and a controller for determining the Asian language of the script text portion of the image, comprising: connected component generating means for determining at least one connected component from the plurality of signal portions; feature determining means for locating script features, corresponding to at least one predetermined type of script feature, for each at least one connected component; feature distribution determining means for determining a distribution of the located script features; comparison means for comparing the determined feature distribution to predetermined feature distributions of at least one Asian language; and output means for outputting a signal indicative of a determined language of the script text portion.

Claim 11

The automatic language determining system of claim 8, wherein the feature determining means comprises: a character cell determining means for determining a plurality of character cells, each one of the plurality of character cells comprising at least one connected component.

Claim 12

The automatic language determining system of claim 9, wherein the feature determining means further comprises feature selecting means for selecting at least one script feature to be located.

Claim 13

The automatic language determining system of claim 9, wherein the at least one predetermined script feature is optical density.

Claim 16

The method of claim 12, further comprising the step of determining a plurality of character cells of the script text portion, each character cell comprising at least one connected component; wherein the at least one preselected type of script feature is optical density of the character cells.