Adaptive color super resolution thresholding
View Patent ↗Methods and systems for adaptively thresholding color image pixels are based on determining a signed (e.g., a plus or minus sign) single-component gray image which can be supersampled and binarized. The method computes the distance of a given color image pixel from an idealized plane which segments the color space into two regions. The process is made adaptive by making the threshold a function of neighborhood pixels. This combines the information from all color channels, such as, for example, 3 or 4 color channels, into a high quality gray channel used for binarization.
1. A method of forming a gray scale image having a plurality of image pixels, comprising:
selecting a neighborhood of image pixels;
determining color coordinates, including at least one of image pixel luminance, density and brightness, of the image pixels in the neighborhood;
determining a minimum vector and maximum vector for the pixels of the selected neighborhood of the color coordinate which corresponds to one of image pixel luminance, density and brightness;
determining a vector average of the maximum and minimum vectors;
determining a vector difference between the maximum and minimum vectors; and
determining a signed single component pixel using the maximum and minimum vectors, the determined vector average and the determined vector difference; and
determining a threshold based on the signed single component gray pixel.
2. The method according to claim 1 , further comprising:
determining system noise; and
testing whether the vector difference is greater than the determined system noise and minimum feature contract limits.
3. The method of claim 2 , further comprising applying static, single component thresholding if the vector difference is not greater than system noise.
4. The method of claim 1 , wherein the color coordinates are specified in one of CIE X,Y,Z; L*,a*,b*; Y*,u″,v*; U*,V*,W*; S, θ, W*; C,M,Y; C,M,Y,K; Y′Cb′Cr; NTSC Y,I,Q; H,S,I; H,S,V; CIE R,G,B; and NTSC R N ,G N ,B N color coordinate systems.
5. A system for forming a gray scale image comprising image pixels, comprising:
a that selects a neighborhood of image pixels;
a color coordinate determiner that determines color coordinates, including at least one of image pixel luminance, density and brightness of the image pixels in the neighborhood;
a vector analyzer that determines a minimum vector and a maximum vector for the pixels in the selected neighborhood of the color coordinate which corresponds to one of image pixel luminance, density and brightness;
a vector averager that determines a vector average of the maximum and minimum vectors;
a vector difference determiner that determines a vector difference between the maximum and minimum vectors;
a gray value determiner that determines a signed single component gray pixel using the maximum and minimum vectors, that determines vector average and the determined vector difference; and
a threshold adapter that determines a threshold based on the signed single-component gray pixel.
6. The method according to claim 5 , further comprising:
a noise analyzer that determines system noise; and
a comparator that compares the vector difference luminance to the system noise and an estimate of the vector difference magnitude to a minimum feature contrast.
7. The system of claim 5 , further comprising a static thresholding unit applied when the comparator that indicates that the vector difference is not greater than the system noise.
8. A method of determining a threshold for thresholding color image values, comprising:
selecting a neighborhood of image pixels;
determining color coordinates, including at least one of image pixel luminance, density and brightness of the image pixels in the neighborhood;
determining a minimum and maximum vectors for the pixels of the selected neighborhood of the color coordinate which corresponds to one of image pixel luminance, density and brightness; and
determining a vector average of the maximum and minimum vectors;
determining a vector difference between the dependent maximum and minimum vectors; and
determining a signed single-component gray pixel using the maximum and minimum vectors, the determined vector average and the determined vector difference; and
determining a threshold based on the signed single-component gray pixel.
9. A system for determining a threshold for thresholding color image pixels, comprising:
a window usable to select a neighborhood of image pixels;
a color coordinate determiner that determines color coordinates, including at least one of image pixel luminance, density and brightness of the image pixels in the neighborhood;
a vector analyzer that determines the dependent minimum and a maximum vector for the neighborhood pixels of the color coordinate which corresponds to one of image pixel luminance, density and brightness;
a computer that determines a signed single-component gray pixel using the maximum and minimum vectors, the determined vector average and the determined vector difference; and
a threshold adapter to determine the threshold based on the signed single-component gray pixel.