Foreground erosion method and system for Multiple Raster Content (MRC) representation of documents
View Patent ↗A foreground erosion method is used to meet thin (but unbroken) text requirements using linear YCC segmentation. The method uses a fixed value subtracted from a gray selector signal, thereby thinning/eroding the foreground. In cases where the pixel would be converted from foreground to background, this is only done if a neighborhood test verifies that the thinning won't result in broken lines. The foreground erosion method attempts to match a couple of templates wherein if a match is found, then the adjustment is performed.
1. A method for thinning or eroding a gray image of an input image defined by a plurality of pixels that preserves topological features when the gray image is binarized comprising:
conducting a test of a neighborhood of the plurality of pixels, the neighborhood comprising an at least 3×3 window of pixels having a center pixel;
subtracting a fixed value from the gray image of the center pixel only if the neighborhood test verifies that the thinning/erosion does not break thin features in a resulting binary output image, wherein
the subtracting is performed if the subtracting causes the central pixel to be converted from a foreground to a background, and any of a plurality of templates match the at least 3×3 window of pixels.
2. The method according to claim 1 wherein the subtracting is also performed only if the neighborhood test verifies that the thinning/erosion does not break thin features in a resulting binary output image, and the central pixel would not be converted from a foreground to a background.
3. The method according to claim 1 , wherein the plurality of templates comprises two templates, each of the two templates having four rotated orientations.
4. The method according to claim 1 , wherein one of two at least 3×3 templates match.
5. The method according to claim 1 , wherein one of the plurality of templates is a 3×3 template that comprises three strong backgrounds along one side and one strong foreground in a middle of the opposite side.
6. The method according to claim 1 , wherein one of the plurality of templates is a 3×3 template that comprises a strong background in a corner of a first side and a second side, a strong background in a middle of the first side, a strong background in a middle of the second side, a strong foreground in a middle of a third side, and a strong foreground in a middle of a fourth side.
7. The method according to claim 3 , wherein the test is performed by comparing the at least 3×3 window with all four rotated orientations of the two templates.
8. A system for thinning or eroding a gray image of an input image defined by a plurality of pixels that preserves topological features when the gray image is binarized, comprising:
a module that tests a neighborhood of the plurality of pixels, the neighborhood comprising an at least 3×3 window of pixels having a center pixel;
a module that subtracts a fixed value from the gray image of the center pixel only if the neighborhood test verifies that the thinning/erosion does not break thin features in a resulting output binary image, wherein
the module subtracts the fixed value if it causes the central pixel to be converted from a foreground to a background, and any of a plurality of templates match the at least 3×3 window of pixels.
9. The system according to claim 8 , wherein the subtracts also is performed only if the neighborhood test verifies that the thinning/erosion does not break thin features in a resulting binary output image, and the central pixel would not be converted from a foreground to a background.
10. The system according to claim 8 , wherein one of the plurality of templates is a 3×3 template that comprises three strong backgrounds along one side and one strong foreground in a middle of the opposite side.
11. The system according to claim 8 , wherein one of the templates is a 3×3 template that comprises a strong background in a corner of a first side and a second side, a strong background in a middle of the first side, a strong background in a middle of the second side, a strong foreground in a middle of a third side, and a strong foreground in a middle of a fourth side.
12. The system according to claim 8 , wherein the test comprises comparing the at least 3×3 window with four rotated orientations of two of the templates.
13. A method for thinning or eroding a gray image of an input image defined by a plurality of pixels that preserves topological features when the gray image is binarized comprising:
conducting a test of a neighborhood of the plurality of pixels, the neighborhood comprising an at least 3×3 window of pixels having a center pixel; and
subtracting a fixed value from the gray image only if the neighborhood test verifies that the thinning/erosion does not break thin features in a resulting binary output image, wherein
the subtracting is performed if the subtracting does not cause the central pixel to be converted from a foreground to a background; and
the subtracting also is performed if the subtracting causes the central pixel to be converted from a foreground to a background and either of two templates match the at least 3×3 window of pixels.
14. The method according to claim 13 , wherein one of the templates is a 3×3 template that comprises three strong backgrounds along one side and one strong foreground in a middle of the opposite side.
15. The method according to claim 14 , wherein one of the templates is a 3×3 template that comprises a strong background in a corner of a first side and a second side, a strong background in a middle of the first side, a strong background in a middle of the second side, a strong foreground in a middle of a third side, and a strong foreground in a middle of a fourth side.
16. The method according to claim 15 , wherein the test is performed by comparing the at least 3×3 window with four rotated orientations of the two of the templates.