Method and its apparatus for classifying defects

• US 20040028276A1
• Filed: 02/28/2003
• Published: 02/12/2004
• Est. Priority Date: 08/12/2002
• Status: Active Grant

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Okuda, Hirohito, Honda, Toshifumi, Takagi, Yuji, Miyamoto, Atsushi
• Granted Patent: US 7,583,832 B2
• Time in Patent Office: Days
• US Class Current: 382/224
• CPC Class Codes: G06T 7/001 using an image reference approach

Abstract

A highly reliable defect automatic classifying method and apparatus capable of flexibly coping with a request for classifying a defect given by each user without having to collect lots of teach data items. A classifying class arrangement is defined by a user by combining classes supplied by the system itself or classes defined by the user and, further, a priori knowledge on the defect class is given by the user as a restriction so as to carry out restricted learning.

Claims

1. A method for classifying defects, comprising the steps of:

imaging a sample for obtaining an image of said sample;

extracting a defect image of said sample by comparing said sample image with a reference image;

associating said extracted defect image with a plurality of corresponding classes for classifying said defect image;

storing information on said association;

specifying a restricting condition for classifying said defect image into said plurality of classes;

classifying said extracted defect image into said plurality of classes by using said stored associating information and said specified restricting condition; and

displaying said classified defect image on a screen.

2. A method for classifying defects as claimed in claim 1, wherein said plurality of classes into which said defect image is to be classified are made by combining preset classes.

3. A method for classifying defects as claimed in claim 1, wherein said associating information to be stored is a feature amount of said defect image corresponding to each of said plurality of classes.

4. A method for classifying defects, comprising the steps of:

imaging a sample for obtaining an image of said sample;

extracting a defect image of said sample by comparing said sample image with a reference image;

specifying a class arrangement composed of plural classes into which said defect image is to be classified;

associating each class of said specified class arrangement with said defect image;

storing associating relation between said specified class arrangement and said defect image associated with each class of said class arrangement;

specifying a restricting condition for classifying said defect image into each class of said specified class arrangement;

classifying said extracted defect image into each class of said specified class arrangement by using said stored associating relation and said specified restricting condition; and

displaying said classified defect image on a screen.

5. A method for classifying defects as claimed in claim 4, further comprising the step of displaying on said screen said defect image extracted in the step of extracting said defect image, and wherein in said step of associating is executed to sort out said defect image displayed on said screen into respective classes of said specified class arrangement.

6. A method for classifying defects as claimed in claim 4, wherein said class arrangement composed of plural classes into which said defect image is to be classified is composed by combining pre-stored classes into which said defect image is to be classified.

7. A method for classifying defects as claimed in claim 4, wherein information on said associating relation with said stored defect image is a feature amount of said defect image corresponding to respective ones of said classes.

8. A method for classifying defects as claimed in claim 4, wherein said restricting condition relates to a type and a magnitude of the feature amount of said defect image.

9. A method for classifying defects, comprising the steps of:

imaging a sample for obtaining an image of said sample;

extracting a defect image of said sample by comparing said sample image with a reference image;

specifying plural classes into which said extracted defect image is to be classified;

associating said specified plural classes with said displayed defect image;

storing an associating relation between said specified plural classes and said displayed defect image;

specifying a restricting condition for classifying said defect image into said specified plural classes;

specifying a class arrangement based on a combination of said specified classes;

classifying said extracted defect image based on said specified class arrangement by using said stored associating relation and said specified restricting condition; and

displaying said classified defect image on a screen.

10. A method for classifying defects as claimed in claim 9, wherein in the step of specifying said plural classes a class into which said extracted defect image is to be classified is select from pre-stored classes, a new class used for classifying said extracted defect image is added, and said plural classes are specified by using said selected, pre-stored class and said added class.

11. A method for classifying defects as claimed in claim 9, further comprising the step of displaying on said screen said extracted defect image extracted in the step of extracting said defect image, and wherein in said step of associating said defect image displayed on said screen is classified into respective classes of said specified class arrangement.

12. A method for classifying a defect as claimed in claim 9, wherein information on said associating relation with said stored defect image is a feature amount of said defect image corresponding to respective ones of said plural classes.

13. A method for classifying defects as claimed in claim 9, wherein said restricting condition is related to a type and a magnitude of said feature amount of said defect image.

15. A method for classifying defects as claimed in claim 9, wherein said pre-stored classifying data is composed of a plurality of feature amounts of said defect image.

14. A method for classifying defects, comprising the steps of:

imaging a sample for obtaining an image of said sample;

extracting a defect image of said sample by comparing said sample image with a reference image;

specifying an arrangement of classes into which said extracted defect image is to be classified;

editing pre-stored classifying data for classifying a defect image into classes in association with said specified class arrangement;

classifying said extracted defect image into said specified class arrangement by using each piece of said classifying data edited in association with said specified class arrangement; and

displaying said classified defect image on a screen.

16. An apparatus for classifying defects, comprising:

means for obtaining an image of a sample by imaging said sample;

means for extracting a defect image of said sample by comparing said sample image obtained by said image obtaining means with a reference image; and

means for classifying said defect image extracted by said defect image extracting means into a plurality of classes,

wherein said defect classifying means includes;

a storage unit for storing information on an associating relation between a class arrangement composed of plural classes into which said defect image obtained by said image obtaining means is to be classified and said defect image;

a restricting condition specifying unit for specifying a restricting condition for classifying said defect image into a class arrangement composed of plural classes;

a defect classifying unit for classifying said defect image extracted by said defect image extracting means into said class arrangement composed of plural classes by using the information on said associating relation stored in said storage unit and said restricting condition specified by said restricting condition specifying unit; and

a display unit for displaying said defect image classified by said defect classifying unit.

17. An apparatus for classifying defects as claimed in claim 16, wherein said class arrangement composed of plural classes used for classifying said defect image, stored in said storage unit, is composed by combining pre-stored classes into which said defect image is to be classified.

18. An apparatus for classifying defects as claimed in claim 16, wherein the information on said associating relation with said defect, stored in said storage unit, is a feature amount of said defect image corresponding to respective ones of said plural classes of said class arrangement.

19. An apparatus for classifying defects as claimed in claim 16, wherein the restricting condition specified by said restricting condition specifying unit is related to a type and a magnitude of a feature amount of said defect image.

20. An apparatus for classifying defects, comprising:

means for imaging a sample for obtaining an image of said sample;

means for extracting a defect image of said sample by comparing said sample image obtained by said imaging means with a reference image;

means for specifying a class arrangement composed of plural classes into which said defect image extracted by said defect extracting means is to be classified;

means for storing classifying data for classifying said defect image into said classes;

means for editing said defect image classifying data stored in said storing means in association with said class arrangement specified by said class arrangement specifying means;

means for classifying said defect image extracted by said defect extracting means into said class arrangement specified by said class arrangement specifying means by using each piece of said classifying data edited by said classifying data editing means; and

means for displaying said defect image classified by said classifying means on a screen.

21. An apparatus for classifying defects as claimed in claim 20, wherein said classifying data stored in said storing means is composed of a plurality of feature amounts of said defect image.