System, a method and a computer program product for fitting based defect detection

Patent metadata

Field	Value
Publication number	US-10290092-B2
Application number	US-201414279192-A
Country	US
Kind code	B2
Filing date	May 15, 2014
Priority date	May 15, 2014
Publication date	May 14, 2019
Grant date	May 14, 2019

Abstract

A system configured to detect defects in an inspection image generated by collecting signals arriving from an article, the system comprising a tangible processor which includes: (i) a distribution acquisition module, configured to acquire a distribution of comparison values, each of the comparison values being indicative of a relationship between a value associated with a pixel of the inspection image and a corresponding reference value; (ii) a fitting module, configured to fit to the distribution an approximation function out of a predefined group of functions; and (iii) a defect detection module, configured to: (a) set a defect detection criterion based on a result of the fitting; and to (b) determine a presence of a defect in the inspection image, based on the defect detection criterion.

First claim

What is claimed is:

1. A method for computerized defect detection in an inspection image generated by collecting signals arriving from an article, the method comprising:
acquiring a distribution of comparison values, each of the comparison values being indicative of a relationship between a value associated with a pixel of the inspection image and a corresponding reference value;
selecting a family of functions, from a plurality of families of functions, based on a noise characteristic of an inspection of the article and a type of defect to be detected;
selecting an approximation function from the selected family of functions;
fitting, to the distribution, the approximation function that is selected from the family of functions;
setting a defect detection criterion based on a result of the fitting; and determining, by a processing device, a presence of a defect in the inspection image based on the defect detection criterion.

2. The method according to claim 1 , wherein the approximation function is a Parabola and the fitting includes selecting one or more Parabola parameters.

3. The method according to claim 1 , wherein the fitting includes selecting the approximation function by estimating errors between the distribution of comparison values and the approximation function.

4. The method according to claim 1 , wherein each of the comparison values is indicative of a difference between a color value of a pixel of the inspection image and a color value of a corresponding pixel of a reference image.

5. The method according to claim 1 , wherein the fitting gives smaller weight to common comparison values.

6. The method according to claim 1 , wherein the acquiring comprises applying an operator to values of the distribution, wherein for at least half of the values of the distribution, a difference between an output value of the operator and a maximum function is less than 10% of a maximum value of the distribution, wherein the maximum function is based on another output value of a logarithmic function and a constant value.

7. The method according to claim 1 , wherein the fitting gives smaller weight to comparison values whose distance from an average comparison value is larger than a standard deviation of the distribution.

8. The method according to claim 1 , wherein the fitting gives smaller weight to: (a) common comparison values, and to (b) comparison values whose distance from an average comparison value is larger than a standard deviation of the distribution.

9. A system configured to detect defects in an inspection image generated by collecting signals arriving from an article, the system comprising:
a memory; and
a processor, operatively coupled with the memory, to:
acquire a distribution of comparison values, each of the comparison values being indicative of a relationship between a value associated with a pixel of the inspection image and a corresponding reference value;
select a family of functions, from a plurality of families of functions, based on a noise characteristic of an inspection of the article and a type of defect to be detected;
select an approximation function from the selected family of functions;
fit, to the distribution, the approximation function that is selected from the family of functions;
set a defect detection criterion based on a result of the fitting; and determine a presence of a defect in the inspection image based on the defect detection criterion.

10. The system according to claim 9 , wherein the approximation function is a Parabola and the fitting includes selecting one or more Parabola parameters.

11. The system according to claim 9 , wherein the processor is further to select the approximation function by estimating errors between the distribution of comparison values and the approximation function.

12. The system according to claim 9 , wherein each of the comparison values is indicative of a difference between a color value of a pixel of the inspection image and a color value of a corresponding pixel of a reference image.

13. The system according to claim 9 , wherein the processor is further to give smaller weight to common comparison values when fitting the approximation function to the distribution.

14. The system according to claim 9 , wherein the processor is further to apply an operator to values of the distribution, wherein for at least half of the values of the distribution, a difference between an output value of the operator and a maximum function is less than 10% of a maximum value of the distribution, wherein the maximum function is based on another output value of a logarithmic function and a constant value.

15. The system according to claim 9 , wherein the processor is further to give smaller weight to comparison values whose distance from an average comparison value is larger than a standard deviation of the distribution when fitting the approximation function to the distribution.

16. The system according to claim 9 , wherein the processor is further to give smaller weight to: (a) common comparison values, and to (b) comparison values whose distance from an average comparison value is larger than a standard deviation of the distribution, when fitting the approximation function to the distribution.

17. A non-transitory computer readable storage medium having instructions that, when executed by a processing device, cause the processing device to perform operations for defect detection in an inspection image generated by collecting signals arriving from an article, the operations comprising:
acquiring a distribution of comparison values, each of the comparison values being indicative of a relationship between a value associated with a pixel of the inspection image and a corresponding reference value;
selecting a family of functions, from a plurality of families of functions, based on a noise characteristic of an inspection of the article and a type of defect to be detected;
selecting an approximation function from the selected family of functions;
fitting, to the distribution, the approximation function that is selected from the family of functions;
setting a defect detection criterion based on a result of the fitting; and determining a presence of a defect in the inspection image based on the defect detection criterion.

18. The non-transitory computer readable storage medium according to claim 17 , wherein the approximation function is a Parabola and the fitting includes selecting one or more Parabola parameters.

19. The non-transitory computer readable storage medium according to claim 17 , wherein the fitting includes selecting the approximation function by estimating errors between the distribution of comparison values and the approximation function.

20. The non-transitory computer readable storage medium according to claim 17 , wherein the fitting gives smaller weight to common comparison values.

21. The non-transitory computer readable storage medium according to claim 17 , wherein the acquiring comprises applying an operator to values of the distribution,
wherein for at least half of the values of the distribution, a difference between an output value of the operator and a maximum function is less than 10% of a maximum value of the distribution, wherein the maximum function is based on another output value of a logarithmic function and a constant value.

22. The non-transitory computer readable storage medium according to claim 17 , wherein the fitting gives smaller weight to comparison values whose distance fr

Assignees

• Applied Materials Israel Ltd

Inventors

• Amzaleg Moshe
• Dodzin Nir Ben-David

Classifications

• G06T2207/20076

  Probabilistic image processing · CPC title

• G06T2207/30148

  Semiconductor; IC; Wafer · CPC title

• G06T7/0008

  checking presence/absence · CPC title

• G06T7/0002

  Inspection of images, e.g. flaw detection · CPC title

• G06T7/0006

  using a design-rule based approach · CPC title