Robust recognition of clusters of streaks at multiple scales
View Patent ↗The present disclosure is generally related to the field of printing, and other processes based on the continuous deposition of material, including industrial molding or extrusion processes, and more particularly to techniques for grouping object quality anomalies in a specified region of an object based on context-specific characterization and shared descriptive profile features.
1. A method of grouping image quality anomalies of an image, the method comprising:
a) printing a test image onto a printable media according to an input characterization data set using a printing system;
(b) measuring the test image to generate a characterization data set representing measurements of intensity or color values observed from the test image;
(c) generating a density profile using the characterization data set, the density profile representing intensity or color value variation of the test image in a cross-process direction transverse to a process direction of the printing system;
(d) operating a processor to determine a descriptor parameter set for a streak template that approximates the density profile using a basis selection algorithm, the descriptor parameter set including:
a location parameter, l, in the cross-process direction,
a width parameter, w, in the cross-process direction, and
an intensity parameter, h, at the location;
(e) updating the density profile according to the streak template and the descriptor parameter set;
(f) generating a printer streak characterization according to the streak template and the descriptor parameter sets;
(g) based on the printer streak characterization, selecting a region of the image to be analyzed, the region defining an absolute cut-off distance within which a streak must be located to be a part of a grouping;
(h) selecting a streak within the region against which all streaks will be compared; and
(i) analyzing all remaining streaks within the region using a transitive closure operation, such that streaks that meet the following criteria for determining similarity can be grouped, the criteria consisting of:
1) location of a streak within a predefined relative distance of the streak selected in step (h); and
2) similarity of the streak from step (i)(1) as compared to the streak selected in step (h) based on the descriptor parameter set determined in step (d);
wherein the streaks are then grouped according to the following rule:
| l 1 −l 2 |≦min( αw 1 ), wherein (α w 1 ) is the cut-off distance
|log h 1 /h 2 |≦β
|log w 1 /w 2 |≦β.
2. The method of claim 1 , wherein clusters represent similar streaks in the process direction of the image.
3. The method of claim 1 , wherein the relative distance is from about 2.25 to about 3.0 times the width of the streak selected in step (h).
4. The method of claim 1 , wherein clusters are identified in fractal fashion.
5. The method of claim 1 , wherein a single image may contain clusters of different scale.
6. The method of claim 1 , including the further step of using the results of step (i) to support diagnostic applications.
7. The method of claim 1 , wherein the descriptor parameter set is determined using a matching pursuit basis selection algorithm.
8. The method of claim 1 , where generating the density profile comprises averaging or integrating the characterization data set.
9. The method of claim 1 , where the test image is printed according to an input characterization data set representing a uniform selected color or gray scale shade.
10. A printer streak characterization system for characterizing and grouping printer streaks, the system comprising:
a printing system including a memory and a processor;
the memory storing a characterization data set representing measurements of intensity or color values observed from an image printed in a printable media;
the processor operative to generate a density profile using the characterization data set, the density profile representing intensity or color value variation of the image in a cross-process direction transverse to a process direction of the printing system;
the processor being operative to determine a descriptor parameter set for a streak template that approximates the density profile using a basis selection algorithm, the descriptor parameter set including:
a location parameter in the cross-process direction,
a width parameter in the cross-process direction, and
an intensity parameter at the location;
the processor being operative to update the density profile according to the streak template and the descriptor parameter set;
the processor being operative to generate a printer streak characterization according to the streak template and the descriptor parameter sets;
the processor being operative to compare streaks within a relative distance of each other on the image based on the descriptor set of each streak that has been characterized; and
the processor being operative to group similar streaks.
11. The printer streak characterization system of claim 10 , where the memory and the processor are integrated into the printing system, the printing system comprising:
at least one print engine operative to print the test image onto a printable media according to an input characterization data set; and
means for measuring the test image to generate the characterization data set representing measurements of intensity or color values observed from the test image.
12. The printer streak characterization system of claim 10 , where the processor is operative to determine the descriptor parameter set using a matching pursuit basis selection algorithm.
13. The printer streak characterization system of claim 10 , where the processor is operative to determine the descriptor parameter set for a streak template by performing a discrete wavelet decomposition using the density profile and a wavelet function to generate:
a location search range for the location parameter along the cross-process direction; and
a width parameter search range for the width parameter along the cross-process direction;
where the processor is operative to restrict the basis selection algorithm according to the location search range and the width parameter search range.
14. The printer streak characterization system of claim 10 , where the test image is printed according to an input characterization data set representing a uniform selected color or gray scale shade.
15. A method of grouping defects in a processed object, the method comprising:
a) processing the object according to an input characterization data set using a processing system;
(b) measuring the object to generate a characterization data set representing measurements of intensity or color values observed from the object;
(c) generating a density profile using the characterization data set, the density profile representing intensity or color value variation of the object in a cross-process direction transverse to a process direction of the processing system;
(d) operating a processor to determine a descriptor parameter set for a streak template that best approximates the density profile using a basis selection algorithm, the descriptor parameter set including:
a location parameter in the cross-process direction,
a width parameter in the cross-process direction, and
an intensity parameter at the location;
(e) updating the density profile according to the streak template and the descriptor parameter set;
(f) generating a streak characterization according to the streak template and the descriptor parameter set;
(g) based on the streak characterization, selecting a region of the object to be analyzed, the region defining an absolute cut-off distance within which a streak must be located to be a part of a grouping;
(h) selecting a streak within the region against which all streaks will be compared; and
(i) analyzing all remaining streaks within the region using a transitive closure operation, such that streaks that meet the following criteria for determining similarity can be grouped, the criteria being:
1) location of a streak within a predefined relative distance of the streak selected in step (h); and
2) similarity of the streak of step (i)(1) as compared to the streak selected in step (h) based on the descriptor parameter set determined in step (d);
wherein the streaks are then grouped according to the following rule:
| l 1 −l 2 |≦min( αw 1 ), wherein (α w 1 ) is the cut-off distance
|log h 1 /h 2 |≦β
|log w 1 /w 2 |≦β.
16. The method of claim 15 , wherein the processing system is an extrusion system.
17. The method of claim 15 , wherein the processing system is a molding system.
18. The method of claim 15 , wherein clusters represent similar streaks in the process direction of the object.
19. The method of claim 15 , wherein the relative distance is from about 2.25 to about 3.0 times the width of the streak selected in step (h).
20. The method of claim 15 , wherein clusters are identified in fractal fashion.
21. The method of claim 15 , wherein a single object may contain clusters of different scale.
22. The method of claim 15 , including the further step of using the results of step (i) to support diagnostic applications.