Exploiting color for license plate recognition
View Patent ↗Methods and systems for exploiting color for object recognition. A color gradient for each pixel in a gradient image of an object can be calculated. The gradient image can be binarized to produce an image having white walls around characters and other structures. A connected component analysis operation can be performed with respect to black regions in the image to determine bounding boxes for characters and other pictorial elements in the image and thereafter identify character candidates from the image utilizing character metrics. Non-character colors can then be eliminated from the image utilizing an outlier rejection.
1. A method for exploiting color for object recognition, said method comprising:
identifying at least one region of an image having character-like features;
obtaining an average color of said at least one region after using an outlier rejection in a perceptually uniform color space to eliminate non-character colors associated with said at least one region; and
modifying all portions of said image with colors that are a fixed amount above said average color, after said obtaining said average color of said least one region after using said outlier rejection to eliminate said noncharacter colors.
2. The method of claim 1 wherein said average color comprises an average character color and wherein said perceptually uniform color spaces comprises a L*a*b* color space, wherein L* represents lightness, a* is a magenta axis with 0 being neutral and green being green, and wherein b* is a yellow axis with 0 being neutral and blue being negative.
3. The method of claim 2 wherein said character-like features comprise a pixel count.
4. The method of claim 1 wherein said outlier rejection includes calculating a covariance matrix for a list containing colors of all pixels assumed to be in said character-like features and then calculating a distance of each pixel among said pixels from said average color using an equation of D=V t M V where V=P−A, V t is a transpose, M is an inverse of said covariance matrix, P is a pixel color, and A is an average character color, and further includes eliminating all pixels where D is greater than a specified value, and then recalculating said covariance matrix and said average color, and iterating until no more pixels are removed.
5. The method of claim 4 wherein said average color comprises an average character color and wherein said character-like features comprise a pixel count and a pixel size.
6. The method of claim 2 wherein said image comprises a three-channel text image.
7. The method of claim 6 further comprising converting said three-channel text image to a single channel image having a contrast between text and background maximized in said single channel image.
8. A system for exploiting color for object recognition, said system comprising:
at least one processor; and
a non-transitory computer-usable medium embodying computer program code, said computer-usable medium capable of communicating with said at least one processor, said computer program code comprising instructions executable by said at least one processor and configured for:
identifying at least one region of an image having character-like features;
obtaining an average color of said at least one region after using an outlier rejection in a perceptually uniform color space to eliminate non-character colors associated with said at least one region; and
modifying all portions of said image with colors that are a fixed amount above said average color, after said obtaining said average color of said least one region after using said outlier rejection to eliminate said noncharacter colors.
9. The system of claim 8 wherein said average color comprises an average character color wherein said perceptually uniform color spaces comprises a L*a*b* color space, and wherein L* represents lightness, a* is a magenta axis with 0 being neutral and green being green, and wherein b* is a yellow axis with 0 being neutral and blue being negative.
10. The system of claim 9 wherein said character-like features comprise a pixel count.
11. The system of claim 8 wherein said outlier rejection includes instructions for calculating a covariance matrix for a list containing colors of all pixels assumed to be in said character-like features and then calculating a distance of each pixel among said pixels from said average color using an equation of D= t M V where V=P−A, V t is a transpose, M is an inverse of said covariance matrix, P is a pixel color, and A is an average character color, and further includes eliminating all pixels where D is greater than a specified value, and then recalculating said covariance matrix and said average color, and iterating until no more pixels are removed.
12. The system of claim 11 wherein said average color comprises an average character color and wherein said character-like features comprise a pixel count and a pixel size.
13. The system of claim 9 wherein said image comprises a three-channel text image.
14. The system of claim 13 wherein said instructions are further configured for converting said three-channel text image to a single channel image having a contrast between text and background maximized in said single channel image.
15. A non-transitory processor-readable medium storing code representing instructions to cause a process for exploiting color for object recognition, said code comprising code to:
identify at least one region of an image having character-like features;
obtain an average color of said at least one region after using an outlier rejection in a perceptually uniform color space to eliminate non-character colors associated with said at least one region; and
modify all portions of said image with colors that are a fixed amount above said average color, after said obtaining said average color of said least one region after using said outlier rejection to eliminate said noncharacter colors.
16. The processor-readable medium of claim 15 wherein said average color comprises an average character color and wherein said perceptually uniform color spaces comprises a L*a*b* color space, wherein L* represents lightness, a* is a magenta axis with 0 being neutral and green being green, and wherein b* is a yellow axis with 0 being neutral and blue being negative.
17. The processor-readable medium of claim 16 wherein said character-like features comprise a pixel count.
18. The processor-readable medium of claim 15 wherein said outlier rejection includes calculating a covariance matrix for a list containing colors of all pixels assumed to be in said character-like features and then calculating a distance of each pixel among said pixels from said average color using an equation of D=V t M V where V=P−A, V t is a transpose, M is an inverse of said covariance matrix, P is a pixel color, and A is an average character color, and further includes eliminating all pixels where D is greater than a specified value, and then recalculating said covariance matrix and said average color, and iterating until no more pixels are removed.
19. The processor-readable medium of claim 18 wherein said average color comprises an average character color and wherein said character-like features comprise a pixel count and a pixel size.
20. The processor-readable medium of claim 16 wherein said image comprises a three-channel text image and wherein said code further comprises code to convert said three-channel text image to a single channel image having a contrast between text and background maximized in said single channel image.