Automatic extraction of text regions and region borders for an electronic work surface

A method for generating an enclosure blob array from a text regions array, the method comprising the steps of:

• for each white pixel in the text region array:
• computing a distance to a Euclidean nearest black pixel in the text regions array;
• and
• setting a corresponding pixel in the text regions array to black if the distance is less than a predetermined multiple of an associated nearest neighbor distance.

A method for extracting a structured border array from a text regions array, the method comprising the steps of:

• defining horizontal bands of white space in the structured border array bounded to the top and bottom by distinct black connected components or image borders in the text regions array;
• labeling each horizontal band of white space in the structured border array with a unique identifier and a length which is a mode of horizontal run lengths in the horizontal band;
• suppressing each horizontal band of white space in the structured border array if the horizontal band is adjacent to a longer horizontal band of white space;
• identifying horizontal border bands of white space in the structured border array;
• extracting a midline in the structured border array for each horizontal band as a horizontal midline of its bounding rectangle;
• defining vertical bands of white space in the structured border array bounded to the left and right by distinct black connected components or image borders in the text regions array;
• labeling each vertical band of white space in the structured border array with a unique identifier and a length which is a mode of vertical run lengths in the vertical band;
• suppressing each vertical band of white space in the structured border array if the vertical band is adjacent to a longer vertical band of white space;
• identifying vertical border bands of white space in the structured border array;
• extracting a midline in the structured border array for each vertical band as a vertical midline of its bounding rectangle;
• detecting all junctions between horizontal and vertical midlines in the structured border array;
• and
• deleting all midline segments between a free termination and a junction in the structured border array.

A method for extracting a freeform linear border array from a text regions array, the method comprising the steps of:

• for each white pixel in the text regions array,
• computing a distance to its Euclidean nearest black pixel in the text regions array;
• labeling the white pixel in the text regions array with a distinct identifier of a black connected component containing the nearest black pixel;
• locating all white pixels in the text regions array having an adjacent white pixel with an equal distance but a different distinct identifier of a black connected component;
• and
• marking a pixel in the freeform linear border array for each pixel located in the locating step.

A computer readable storage medium comprising:

• computer readable program code embodied on said computer readable storage medium, said computer readable program code for programming a computer to perform a method for generating a text lines array from an input image array, the method comprising the steps of:
• computing a sufficient stability threshold;
• and
• for each white pixel in the input image array, setting a corresponding pixel in the text lines array to black if a horizontal distance value from the white pixel to a nearest black pixel in a same image pixel row is less than the sufficient stability threshold.

A computer readable storage medium comprising:

• computer readable program code embodied on said computer readable storage medium, said computer readable program code for programming a computer to perform a method for generating a text regions array from a text lines array, the method comprising the steps of:
• computing a maximum frequency stability threshold;
• and
• for each white pixel in the text lines array, setting a corresponding pixel in the text regions array to black if a vertical distance value is not greater than the maximum frequency stability threshold.

A computer readable storage medium comprising:

• computer readable program code embodied on said computer readable storage medium, said computer readable program code for programming a computer to perform a method for generating an enclosure blob array from a text regions array, the method comprising the steps of:
• for each white pixel in the text region array:
• computing a distance to a Euclidean nearest black pixel in the text regions array;
• and
• setting a corresponding pixel in the text regions array to black if the distance is less than a predetermined multiple of an associated nearest neighbor distance.

A computer readable storage medium comprising:

• computer readable program code embodied on said computer readable storage medium, said computer readable program code for programming a computer to perform a method for extracting a structured border array from a text regions array, the method comprising the steps of:
• defining horizontal bands of white space in the structured border array bounded to the top and bottom by distinct black connected components or image borders in the text regions array;
• labeling each horizontal band of white space in the structured border array with a unique identifier and a length which is a mode of horizontal run lengths in the horizontal band;
• suppressing each horizontal band of white space in the structured border array if the horizontal band is adjacent to a longer horizontal band of white space;
• identifying horizontal border bands of white space in the structured border array;
• extracting a midline in the structured border array for each horizontal band as a horizontal midline of its bounding rectangle;
• defining vertical bands of white space in the structured border array bounded to the left and right by distinct black connected components or image borders in the text regions array;
• labeling each vertical band of white space in the structured border array with a unique identifier and a length which is a mode of vertical run lengths in the vertical band;
• suppressing each vertical band of white space in the structured border array if the vertical band is adjacent to a longer vertical band of white space;
• identifying vertical border bands of white space in the structured border array;
• extracting a midline in the structured border array for each vertical band as a vertical midline of its bounding rectangle;
• detecting all junctions between horizontal and vertical midlines in the structured border array;
• and
• deleting all midline segments between a free termination and a junction in the structured border array.

A computer readable storage medium comprising:

• computer readable program code embodied on said computer readable storage medium, said computer readable program code for programming a computer to perform a method for extracting a freeform linear border array from a text regions array, the method comprising the steps of:
• for each white pixel in the text regions array,
• computing a distance to its Euclidean nearest black pixel in the text regions array;
• labeling the white pixel in the text regions array with a distinct identifier of a black connected component containing the nearest black pixel;
• locating all white pixels in the text regions array having an adjacent white pixel with an equal distance but a different distinct identifier of a black connected component;
• and
• marking a pixel in the freeform linear border array for each pixel located in the locating step.