Image binarization for optical character recognition
View Patent ↗Systems and methods convert to binary an input image having pixels defining text and background. Thresholds are determined by which pixels in the input image and a corresponding blurred image will be defined as either binary black or binary white. Thresholds derive from grouped together neighboring pixels having pixels separated out that correspond to the background. For pixels of the input image defined as binary black and having corresponding pixels in the blurred image defined as binary black relative to their thresholds, those are set to black in the binary image, else they are set white. Techniques for devising thresholds, blurring images, grouping together pixels, statistical analysis, etc., typify the embodiments.
1. A method of converting to a binary image an input image having pluralities of pixels defining text and background each with an intensity value defined on a scale of multiple intensity values, comprising:
blurring the input image so that the pixels of the input image have corresponding blurred pluralities of pixels each with blurred intensity values defined on said scale of multiple intensity values;
comparing the intensity value of a pixel under consideration of the input image to a first threshold value defined on the scale of multiple intensity values, the first threshold value being devised from pixels neighboring said pixel under consideration; and
comparing the blurred intensity value of a blurred version of the pixel under consideration of the input image to a second threshold value defined on the scale of multiple intensity values, the second threshold value being devised from pixels neighboring the blurred version of the pixel under consideration,
wherein if both the intensity value of the pixel under consideration is less than or equal to the first threshold value and the blurred intensity value is less than or equal to the second threshold value, setting the intensity value of the pixel under consideration to black in the binary image, else setting the intensity value of the pixel under consideration to white in the binary image.
2. A method of converting to a binary image an input image having pluralities of pixels defining text and background each with an intensity value defined on a scale of multiple intensity values, comprising:
comparing the intensity value of a pixel under consideration of the input image to a first threshold value defined on the scale of multiple intensity values, the first threshold value being devised from pixels neighboring said pixel under consideration; and
comparing a blurred intensity value of a blurred version of the pixel under consideration of the input image to a second threshold value defined on the scale of multiple intensity values, the second threshold value being devised from pixels neighboring the blurred version of the pixel under consideration,
wherein if both the intensity value of the pixel under consideration is less than or equal to the first threshold value and the blurred intensity value is less than or equal to the second threshold value, setting the intensity value of the pixel under consideration to black in the binary image, else setting the intensity value of the pixel under consideration to white in the binary image.
3. The method of claim 2 , further including blurring the input image such that the pixel under consideration of the input image has a corresponding blurred pixel under consideration in the blurred image.
4. A method of binarization of an input image having pluralities of pixels defining text and background, comprising:
blurring the input image to obtain a blurred image defining blurred text and background;
for both the input image and the blurred image, grouping together pluralities of neighboring pixels;
identifying pixels that represent the background and blurred background in the grouped together pixels;
separating out remaining pixels from the identified pixels representing the background and blurred background in the grouped together pixels; and
determining thresholds by which the pixels in the input image and corresponding pixels in the blurred image will be defined as either a binary black pixel or binary white pixel.
5. The method of claim 4 , further including setting to black pixels in a binary image representation of the input image those pixels of the input image being defined as said binary black pixels and having corresponding pixels in the blurred image being defined as said binary black pixels relative to the thresholds.
6. A stand-alone or networked imaging device, comprising:
a scanner to scan documents into an input image having pluralities of pixels defining text and background; and
one or more controllers to convert the input image to a binary image by
blurring the input image to obtain a blurred image,
determining thresholds for the pixels of the input image and corresponding pixels in the blurred image by which the pixels and corresponding pixels will be defined as either a black pixel or a white pixel, and
setting to black pixels in the binary image those pixels of the input image being defined as said black pixels and having corresponding pixels in the blurred image being defined as said black pixels relative to the thresholds.
7. The imaging device of claim 6 , further including an optical character recognition (OCR) engine to take the binary image as an input to an OCR algorithm.