Line region segmentation method

Claim 2

A line region segmentation method comprising the steps of: grouping pixels of an input image described by image signals into three sets of black, gray and white pixels; performing a first pattern matching of black pixels included in a given region of said input image with a first set of predetermined matrix patterns on the basis of said image signals to determine whether a black pixel among said set of black pixels is closely associated with other pixels of said set of black pixels for each pixel of said set of black pixels; performing a second pattern matching of black pixels included in a given region of said input image with a second set of predetermined matrix patterns on the basis of said image signals to determine whether a black pixel among said set of black pixels is one of screened black pixels in said input image for each pixel of said set of black pixels; making a determination as to whether the number of black pixels included in a given region that are determined as being associated black pixels and determined as being non-screened black pixels is greater than a predetermined reference value, so that it is decided that a black pixel being considered in said given region is one of black bilevel pixels when said number of black pixels is greater than said predetermined reference value; performing a third pattern matching of white pixels included in a given region of said input image with a third set of predetermined matrix patterns on the basis of said image signals to determine whether a white pixel among said set of white pixels is closely associated with other pixels of said set of white pixels for each pixel of said set of white pixels; making a determination as to whether the number of white pixels included in a given region that are determined as being associated white pixels is greater than a predetermined reference value, so that it is decided that a white pixel being considered in said given region is one of white bilevel pixels when said number of white pixels is greater than said predetermined reference value; and making a line region discrimination by determining whether a pixel being considered in a given block of said input image is one of bilevel pixels by checking if the number of the thus decided black bilevel pixels included in said given block is greater than a predetermined number and if the number of the thus decided white bilevel pixels included in said given block is greater than a predetermined number.

Claim 3

The method as claimed in claim 1, further comprising the step of performing a dilatation after said line region discrimination is made, so that line regions including not only edge pixels located at edge portions of bilevel images but also internal pixels located inside said bilevel images are segmented from said input image.

Claim 4

The method as claimed in claim 1, wherein said three sets of black, gray and white pixels are supplied from two comparators having respectively high and low threshold values which are predetermined, said three sets being supplied after an optical density of each of the pixels of said input image is compared with said high threshold value of one of said two comparators and the same is compared with said low threshold value of the other comparator.

Claim 5

The method as claimed in claim 3, wherein said set of gray pixels are eliminated from said pixels of said input image for which said pattern matching is performed, each pixel of said set of gray pixels having an intermediate optical density between said two threshold values, and no pattern matching being performed for said set of gray pixels, thus reducing influence of noises included in said image signals.

Claim 6

The method as claimed in claim 1, wherein each matrix pattern of said second set of predetermined matrix patterns have an area factor of black pixels which is approximately equal to 50 percent.

Claim 7

A line region segmentation method comprising the steps of: grouping pixels of an input image described by image signals into three sets of black, gray and white pixels; performing a first pattern matching of black pixels included in a given region of said input image with a first set of predetermined matrix patterns on the basis of said image signals to determine whether a black pixel among said set of black pixels is closely associated with other pixels of said set of black pixels for each pixel of said set of black pixels; performing a second pattern matching of black pixels included in a given region of said input image with a second set of predetermined matrix patterns on the basis of said image signals to determine whether a black pixel among said set of black pixels is one of screened black pixels in said input image for each pixel of said set of black pixels; making a determination as to whether the number of black pixels included in a given region that are determined as being associated black pixels and determined as being non-screened black pixels is greater than a predetermined reference value, so that it is decided that a black pixel being considered in said given region is a black bilevel pixel when said number of black pixels is greater than said predetermined reference value; performing a third pattern matching of white pixels included in a given region of said input image with a third set of predetermined matrix patterns on the basis of said image signals to determine whether a white pixel among said set of white pixels is closely associated with other pixels of said set of white pixels for each pixel of said set of white pixels; performing a fourth pattern matching of white pixels included in a given region of said input image with a fourth set of predetermined matrix patterns on the basis of said image signals to determine whether a white pixel among said set of white pixels is one of screened white pixels in said input image for each pixel of said set of white pixels; making a determination as to whether the number of white pixels included in a given region that are determined as one of associated white pixels and determined as one of non-screened white pixels is greater than a predetermined reference value, so that it is decided that a white pixel being considered in said given region is a white bilevel pixel when said number of white pixels is greater than said predetermined reference value; and making a line region discrimination by determining whether a pixel being considered in a given block of said input image is one of bilevel pixels by checking if the number of the thus decided black bilevel pixels included in said given block is greater than a predetermined number and if the number of the thus decided white bilevel pixels included in said given block is greater than a predetermined number.

Claim 8

A line region segmentation method comprising the steps of: grouping pixels of an input image described by image signals into three sets of black, gray and white pixels; performing a first pattern matching of black pixels included in a given region of said input image with a first set of predetermined matrix patterns on the basis of said image signals to determine whether a black pixel among said set of black pixels is closely associated with other pixels of said set of black pixels for each pixel of said set of black pixels; performing a second pattern matching of black pixels included in a given region of said input image with a second set of predetermined matrix patterns on the basis of said image signals to determine whether a black pixel among said set of black pixels is one of screened black pixels in said input image for each pixel of said set of black pixels; making a determination as to whether the number of black pixels included in a given region that are determined as being associated black pixels and determined as being non-screened black pixels is greater than a predetermined reference value, so that it is decided that a black pixel being considered in said given region is a black bilevel pixel when said number of black pixels is greater than said predetermined reference value; performing a third pattern matching of white pixels included in a given region of said input image with a third set of predetermined matrix patterns on the basis of said image signals to determine whether a white pixel among said set of white pixels is closely associated with other pixels of said set of white pixels for each pixel of said set of white pixels; performing a fourth pattern matching of white pixels included in a given region of said input image with a fourth set of predetermined matrix patterns on the basis of said image signals to determine whether a white pixel among said set of white pixels is one of screened white pixels in said input image for each pixel of said set of white pixels; making a determination as to whether the number of white pixels included in a given region that are determined as one of associated white pixels and determined as one of non-screened white pixels is greater than a predetermined reference value, so that it is decided that a white pixel being considered in said given region is a white bilevel pixel when said number of white pixels is greater than said predetermined reference value; and making a line region discrimination by determining whether a pixel being considered in a given block of said input image is one of bilevel pixels by checking if the number of the thus decided black bilevel pixels included in said given block is greater than a predetermined number and if the number of the thus decided white bilevel pixels included in said given block is greater than a predetermined number.

Claim 9

The method as claimed in claim 6, further comprising the step of performing a dilatation after said line region discrimination is made, so that line regions including not only edge portion pixels located at edge portions of bilevel images but also internal pixels located inside said bilevel images are discriminated from said input image.

Claim 10

The method as claimed in claim 6, wherein said three sets of black, gray and white pixels are supplied from two comparators in which high and low thresholds are predetermined by comparing an optical density of each of the pixels of said input image with said high threshold of one of said comparators and comparing the same with said low threshold of the other comparator.

Claim 11

The method as claimed in claim 8, wherein said set of gray pixels are eliminated from said pixels of said input image for which said pattern matching are performed, each pixel of said set of gray pixels having an intermediate optical density between said two threshold values, and no pattern matching being performed for said set of gray pixels, thus allowing influence of noises included in said image signals to be reduced.

Claim 12

The method as claimed in claim 6, wherein each matrix pattern of said second and fourth sets of predetermined matrix patterns have an area factor of black pixels which is approximately equal to 50 percent.