Document segmentation for mixed raster content representation
View Patent ↗In an array of pixels, a method for segmenting a selected pixel of the array between at least two layers including identifying an N-by-N window centered upon the selected pixel, evaluating at least one pixel in the N-by-N window to determine whether the selected pixel is a potential text element, identifying an M-by-M window centered upon the selected pixel when the evaluation determines that the selected pixel is a potential text element, wherein the M-by-M window is smaller than the N-by-N window, and determining whether the potential text element includes text by comparing at least two pixels within the M-by-M window.
1. A method for compressing an array of pixels representing image data, wherein each pixel in said array has an associated intensity value, comprising:
identifying an N-by-N window centered upon a selected pixel of said array of pixels;
evaluating at least one pixel in said N-by-N window to determine whether said selected pixel is a potential text element, wherein said evaluating comprises:
summing said intensity values of each pixel in a row for each row of pixels within said N-by-N window and identifying a row having a maximum total intensity and a row having a minimum total intensity, wherein a difference between said maximum total intensity of said row and said minimum total intensity of said row is compared to a first predetermined threshold value;
summing said intensity values of each pixel in a column for each column of pixels within said N-by-N window and identifying a column having a maximum total intensity and a column having a minimum total intensity, wherein a difference between said maximum total intensity of said column and said minimum total intensity of said column is compared to a second predetermined threshold value; and
summing said intensity values of each pixel in a first major diagonal of said N-by-N window to obtain a first diagonal total intensity, summing said intensity values of each pixel in a second major diagonal of said N-by-N window to obtain a second diagonal total intensity and comparing a difference between said first diagonal total intensity and said second diagonal total intensity against a third predetermined threshold value;
after said identifying of said N-by-N window, if said evaluating determines that said selected pixel is a potential text element, then identifying an M-by-M window centered upon said selected pixel such that said M-by-M window is smaller than said N-by-N window;
determining whether said potential text element includes text by comparing at least two pixels within said M-by-M window;
assigning said selected pixel to a first layer of said plurality of layers if said determining determines that said potential text element includes text;
assigning said selected pixel to a second layer of said plurality of layers if said determining determines that said potential text element does not include text; and
compressing said first layer using a first compression technique to create a first compressed layer.
2. The method of claim 1 , further comprising compressing said second layer using a second compression technique.
3. The method of claim 2 , wherein said first compression technique includes a lossless compression technique.
4. The method of claim 2 , wherein said first compression technique includes LZW.
5. The method of claim 2 , wherein said second compression technique includes a lossy compression technique.
6. The method of claim 2 , wherein said second compression technique includes JPEG.
7. The method of claim 1 , wherein N is a number between about 5 and about 500.
8. The method of claim 1 , wherein M is a number between about 2 and about 300.
9. The method of claim 1 , wherein said determining includes comparing at least two oppositely positioned pixels, wherein said oppositely positioned pixels border said M-by-M window.
10. A method for compressing an array of pixels representing image data into a plurality of layers, comprising:
identifying an N-by-N window centered on said selected pixel;
obtaining a total intensity value for each row, column and major diagonal of said N-by-N window;
comparing said total intensity values to determine whether said selected pixel is a potential text element;
after said identifying of said N-by-N window, if said selected pixel is determined to be a potential text element, then identifying an M-by-M window centered on said selected pixel such that said M-by-M window is smaller than said N-by-N window;
comparing at least two oppositely positioned pixels within said M-by-M window to determine whether said potential text element includes text; and
assigning said selected pixel to a first layer of the plurality of layers if said potential text element includes text;
assigning said selected pixel to a second layer of said plurality of layers if said potential text element does not include text; and
compressing said first layer using a first compression technique to create a first compressed layer.
11. The method of claim 10 , further comprising compressing said second layer using a second compression technique.
12. The method of claim 11 , wherein said first compression technique includes a lossless compression technique.
13. The method of claim 11 , wherein said second compression technique includes a lossy compression technique.