Method and apparatus of pattern recognition

• US 5,568,563 A
• Filed: 05/03/1994
• Issued: 10/22/1996
• Est. Priority Date: 05/17/1993
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Inventors: Shimizu, Masako, Tanaka, Ken-ichi
• Primary Examiner(s): Mancuso, Joseph
• Application Number:
  US08/237,138
• Time in Patent Office: 903 Days
• Field of Search: 382/209, 382/216, 382/181, 382/218, 382/221, 382/293, 382/294, 382/296, 382/144, 382/147, 382/149, 382/236, 382/150, 348/129, 348/130, 348/87
• US Class Current: 382/144
• CPC Class Codes: G06T 2207/10056 Microscopic image
• CPC Class Codes: G06T 2207/30148 Semiconductor; IC; Wafer
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: H01L 22/12 for structural parameters, e.g. thickness, line width, refractive index, temperature, warp, bond strength, defects, optical inspection, electrical measurement of structural dimensions, metallurgic measurement of diffusions electrical measurement of diffusions H01L22/14

Abstract

A pattern recognition apparatus and method in which a set is produced which includes a fundamental pattern vector on a basis place and other fundamental pattern vectors of the patterns displaced from the fundamental pattern on the basis place. Then a subspace spanned by fundamental pattern vectors included in the set is generated. A test pattern vector of a wafer to be inspected is projected to the subspace and similarity between the fundamental vectors and the test pattern vector is measured. Further, an image is used after it is filtered by a normalization filter. Furthermore, sensitivity of pattern recognition is varied by changing the dimension of the pattern vectors. Moreover, for objects expressed by numerical values which can not be compared directly, the data of the objects are transformed into images and, then, a set of fundamental pattern vectors are worked out.

Claims

1. A pattern recognition method for comparing a first image including a fundamental pattern, to a second image including a test pattern, to detect a match between the test pattern and the fundamental pattern, the method comprising the steps of:

producing a first signal representative of said first image;

producing a second signal representative of said second image;

generating a set of data structures that defines a set of vectors based upon said first signal, said set of vectors including a first vector indicative of the fundamental pattern of the first image disposed in a first position and a plurality of second vectors indicative of the fundamental pattern of the first image disposed in a plurality of respective different positions from the first position;

generating a linear subspace data structure that defines a linear subspace spanned by the first vector and the plurality of second vectors;

generating a test pattern data structure that defines a test pattern vector based upon said second signal, said test pattern data structure being indicative of the second image;

projecting the test pattern vector to the generated linear subspace to produce a third vector which belongs to a complementary subspace of the linear subspace;

generating a norm signal based upon the third vector which belongs to the complementary subspace; and

determining if the fundamental pattern of the first image matches the test pattern of the second image based on the norm signal.

2. The pattern recognition method according to claim 1, wherein the plurality of second vectors includes N second vectors, and each of the first vector and the plurality of second vectors has a dimension greater than N+1.

3. The pattern recognition method according to claim 1, wherein the step of determining includes the step of:

comparing the norm signal to a sensitivity threshold, wherein the sensitivity threshold is based upon a number of dimensions of the set of vectors.

4. The pattern recognition method according to claim 1, wherein the fundamental pattern of the first image is an outward appearance of a structure of a flawless wafer, wherein the test pattern of the second image is an outward appearance of a structure of a wafer under test, and wherein the step of determining includes a step of detecting a defect in the structure of the wafer under test.

5. The pattern recognition method according to claim 1, wherein the step of producing the first signal further comprises a step of normalizing the first image so that the first signal represents a normalized first image, and the step of producing the second signal further comprises a step of normalizing the second image so that the second signal represents a normalized second image.

6. The pattern recognition method according to claim 1, wherein the step of generating the set of data structures includes the step of:

providing the plurality of second vectors indicative of the first image having pixels of the fundamental pattern displaced in parallel relative to the first position.

7. The pattern recognition method according to claim 1, wherein the step of generating the set of data structures includes the step of:

providing the plurality of second vectors indicative of the first image having pixels of the fundamental pattern displaced rotationally with respect to the first position.

8. A pattern recognition method for comparing a first image including a fundamental pattern, to a second image including a test pattern, to detect a match between the test pattern and the fundamental pattern, the method comprising the steps of:

producing a first signal representative of said first image;

producing a second signal representative of said second image;

generating a set of data structures that defines a set of vectors based upon said first signal, said set of vectors including a first vector indicative of the first pattern of the first image disposed in a first position and a plurality of second vectors indicative of the first pattern of the first image disposed in a plurality of respective different positions from the first position;

generating a linear subspace data structure that defines a linear subspace spanned by the first vector and the plurality of second vectors;

generating a test pattern data structure that defines a test pattern vector based upon said second signal, said test pattern data structure being indicative of the second image;

projecting the test pattern vector to the generated linear subspace to produce a third vector which belongs to the linear subspace;

generating a norm signal based upon the third vector which belongs to the linear subspace; and

determining if the fundamental pattern of the first image matches the test pattern of the second image based on the norm signal.

9. The pattern recognition method according to claim 8, wherein the step of generating the set of data structures includes the step of;

providing the plurality of second vectors indicative of the first image having pixels of the fundamental pattern displaced in parallel relative to the first position.

10. The pattern recognition method according to claim 8, wherein the step of generating the set of data structures includes the step of:

providing the plurality of second vectors indicative of the first image having pixels of the fundamental pattern displaced rotationally with respect to the first position.

11. A pattern recognition apparatus for inspecting an outward appearance of a test object, comprising:

means for producing a first signal that represents a first image having a fundamental pattern indicative of a fundamental object, the first image being a reference;

means for producing a second signal that represents a second image, the second image having a test pattern indicative of the test object;

means for generating a set of vectors based upon said first signal including a first vector indicative of the fundamental pattern of the first image disposed in a first position and a plurality of second vectors indicative of the fundamental pattern disposed in a plurality of respective different positions from the first position;

means for generating a linear subspace spanned by the first vector and the plurality of second vectors;

means for generating a test pattern vector indicative of the second image based upon said second signal;

means for projecting the test pattern vector to the generated linear subspace to generate a third vector which belongs to a complementary subspace of the linear subspace;

means for generating a norm signal based upon the third vector which belongs to the complementary subspace; and

means for determining if the fundamental pattern of the first image matches the test pattern of the second image based on the norm signal of the third vector.

12. The pattern recognition apparatus according to claim 11, wherein said means for producing said first signal further comprises a first normalizing filter which normalizes the first image, wherein said means for producing said second signal further comprises a second normalizing filter which normalizes the second image, and wherein the first signal and the second signal are generated based on the normalized first image and the normalized second image.

13. A pattern recognition apparatus for inspecting an outward appearance of a test object, comprising:

means for producing a first signal that represents a first image having a fundamental pattern indicative of a fundamental object, the first image being a reference;

means for producing a second signal that represents a second image, the second image having a test pattern indicative of the test object;

means for generating a set of vectors based upon said first signal including a first vector indicative of fundamental pattern of the first image disposed in a first position and a plurality of second vectors indicative of the fundamental pattern disposed in a plurality of respective different positions from the first position;

means for generating a linear subspace sprained by the first vector and the plurality of second vectors;

means for generating a test pattern vector indicative of the second image based upon said second signal;

means for projecting the test pattern vector to the generated linear subspace to generate a third vector which belongs to the linear subspace;

means for generating a norm signal based upon the third vector which belongs to the linear subspace; and

means for determining if the fundamental pattern of the first image matches the test pattern of the second image based on the norm signal of the third vector.

14. The pattern recognition apparatus according to claim 13, wherein the means for generating the set of data structures includes:

means for providing the plurality of second vectors indicative of the first image having pixels of the fundamental pattern displaced in parallel relative to the first position.

15. The pattern recognition apparatus according to claim 13, wherein the means for generating the set of data structures includes:

means for providing the plurality of second vectors indicative of the first image having pixels of the fundamental pattern displaced rotationally with respect to the first position.

16. The pattern recognition apparatus according to claim 15, further comprising an image shifter, coupled to the means for producing the first signal and the means for producing the second signal, which shifts a barycenter of the fundamental pattern of the first image to the first position, and shifts a barycenter of the test pattern of the second image to a predetermined position.