Skew detection of text in a noisy digitized image
AssigneeInternational Business Machines Corporation
InventorYaakov Navon
Life of this patent
Estimating the term…
Abstract
A method for deskewing text is disclosed, in which a noisy digitized gray scale image is preprocessed by binarization, noise removal, and then by smearing in three patterns. A selected smeared image having minimal runs is sampled, and the resulting sampled image subjected to the Hough process using a search technique at two angular resolutions. A dominant text line is chosen from the output of the Hough process, and the image deskewed accordingly. The method is suitable for detecting skew at any angle, and is capable of detecting multiple skew text lines.
Claims27
In a text image processing system, a method for text skew detection, comprising the steps of:
- smearing a digitized image in one of a horizontal direction and a vertical direction to yield a first smeared image;
- smearing the digitized image in another of said horizontal direction and said vertical direction to yield a second smeared image;
- smearing the digitized image in both of said horizontal direction and said vertical direction to yield a third smeared image;
- and
- selecting one of said first, second and third smeared images according to a minimum number of runs therein;
- and
- selecting a dominant line from the selected smeared digitized image, wherein a text skew angle is determined by a direction of the dominant line.
A computer software product, comprising a computer-readable medium in which computer program instructions are stored, which instructions, when read by a computer, cause the computer to accept as input a digitized image having textual material embedded therein, and to determine at least one skew angle of the textual material by the steps of:
- smearing the digitized image in one of a horizontal direction and a vertical direction to yield a first smeared image;
- smearing the digitized image in another of said horizontal direction and said vertical direction to yield a second smeared image;
- smearing the digitized image in both of said horizontal direction and said vertical direction to yield a third smeared image;
- and
- selecting one of said first, second and third smeared images according to a minimum number of runs therein;
- and
- selecting a dominant line from the selected smeared digitized image, wherein a text skew angle is determined by a direction of the dominant line.
An automated classification system, comprising:
- a conveyance for transporting items to be classified, said items having classification information written thereon;
- an acquisition device having a field of view directed toward said conveyance for acquiring an image of said classification information;
- a digitizing unit for converting said acquired image to a digitized image;
- a computer executing a program of instructions which causes the computer to accept as input said digitized image, and to determine at least one skew angle of the classification information by the steps of:
- expansively smearing the digitized image in one of a horizontal direction and a vertical direction to yield a first smeared image;
- smearing the digitized image in another of said horizontal direction and said vertical direction to yield a second smeared image;
- smearing the digitized image in both of said horizontal direction and said vertical direction to yield a third smeared image;
- and
- selecting one of said first, second and third smeared images according to a minimum number of runs therein;
- selecting a dominant line from the selected smeared digitized image, wherein a text skew angle is determined by a direction of the dominant line;
- an analysis program of said computer accepting said digitized image and said skew angle for deskewing said digitized image and evaluating said classification information, wherein control signals are output from said computer responsive to said analysis program;
- and
- a classifier for classifying said transported items responsive to said control signals.
In a text image processing system, a method for text skew detection, comprising the steps of:
- constructing exactly two smeared digitized images by performing any two of the steps of:
- expansively smearing a digitized image in one of a horizontal direction and a vertical direction to yield a first smeared image;
- expansively smearing the digitized image in another of said horizontal direction and said vertical direction to yield a second smeared image;
- and
- expansively smearing the digitized image in both of said horizontal direction and said vertical direction to yield a third smeared image;
- selecting one of the two smeared digitized images according to a minimum number of runs therein;
- and
- selecting a dominant line from the selected smeared digitized image, wherein a text skew angle is determined by a direction of the dominant line.
A computer software product, comprising a computer-readable medium in which computer program instructions are stored, which instructions, when read by a computer, cause the computer to accept as input a digitized image having textual material embedded therein, and to determine at least one skew angle of the textual material by the steps of:
- constructing exactly two smeared digitized images by performing any two of the steps of:
- expansively smearing a digitized image in one of a horizontal direction and a vertical direction to yield a first smeared image;
- expansively smearing the digitized image in another of said horizontal direction and said vertical direction to yield a second smeared image;
- and
- expansively smearing the digitized image in both of said horizontal direction and said vertical direction to yield a third smeared image;
- selecting one of said two smeared digitized images according to a minimum number of runs therein;
- and
- selecting a dominant line from the selected smeared digitized image, wherein a text skew angle is determined by a direction of the dominant line.
An automated classification system, comprising:
- a conveyance for transporting items to be classified, said items having classification information written thereon;
- an acquisition device having a field of view directed toward said conveyance for acquiring an image of said classification information;
- a digitizing unit for converting said acquired image to a digitized image;
- a computer executing a program of instructions which causes the computer to accept as input said digitized image, and to determine at least one skew angle of the classification information by the steps of:
- constructing exactly two smeared digitized images by performing any two of the steps of:
- expansively smearing a digitized image in one of a horizontal direction and a vertical direction to yield a first smeared image;
- expansively smearing the digitized image in another of said horizontal direction and said vertical direction to yield a second smeared image;
- and
- expansively smearing the digitized image in both of said horizontal direction and said vertical direction to yield a third smeared image;
- selecting one of said two smeared digitized images according to a minimum number of runs therein;
- and
- selecting a dominant line from the selected smeared digitized image, wherein a text skew angle is determined by a direction of the dominant line;
- an analysis program of said computer accepting said digitized image and said skew angle for deskewing said digitized image and evaluating said classification information, wherein control signals are output from said computer responsive to said analysis program;
- and
- a classifier for classifying said transported items responsive to said control signals.
Dependencies are inferred from the claim text.