Progressive self-learning defect review and classification method

1. A progressive self-learning (PSL) method for enhancing wafer or mask defect review and classification without reference to a pre-existing defect library to accelerate a defect review and classification process comprising the steps of:

performing image processing on a scanned defect image retrieved from a scanned defect image file to produce an extracted image comprising a defect map representation comprising two basic shapes including a bulky square region and a round region;

aligning the scanned defect image with a just-stored digitized defect image comprising a defect map representation comprising two basic shapes including a bulky square region and a round region;

determining if there is a match between the defect map representation produced from the scanned defect image and the defect map produced from the just-stored digitized defect image; and

classifying the scanned defect image according to a classification associated with the just-stored digitized defect image in the case of a match; or

adding the defect map representation produced from the scanned defect image to a newly formed database comprising the just-stored digitized defect image including manually classifying said 2-bit defect map in the case there is not a match.

2. The PSL method of claim 1 wherein the step of performing image processing on a scanned defect image from a scanned defect image file further comprises the steps of:

extracting a scanned defect image from a scanned defect image file; and

digitizing the scanned defect image using preselected digitizing parameters.

3. The PSL method of claim 2 wherein the step of extracting a scanned defect image comprises the, step of:

extracting a defect region having a dimension of n pixels by n pixels from a scanned image, wherein n is a pre-defined programmable number.

4. The PSL method of claim 2 wherein the step of digitizing comprises the steps of:

converting the extracted image into a 2-bit defect map; and

transforming the 2-bit defect map into a segmented image to produce said defect map representation.

5. The PSL method of claim 4 wherein the step of converting the extracted image into a 2-bit defect map comprises the step of:

using a square tessellation method.

6. The PSL method of claim 5 wherein the step of transforming the 2bit defect map into a segmented image alignment step comprises the step of:

transforming a sub-region of said 2-bit defect map into one or more polygons to produce a defect image shape.

7. The PSL method of claim 1 wherein the bulky square regions comprises a reference to align the bulky square region comprising the scanned defect image with the bulky square region comprising the just-stored digitized defect image.

8. The PSL method of claim 1 wherein the round region comprising the scanned defect image comprises a reference in the alignment step to locate a defect position.

9. The PSL method of claim 7 wherein overlapping regions of said bulky square regions are aligned in accordance with an alignment formula.

10. The PSL method of claim 1 wherein the step of determining if there is a match comprises the step of:

calculating an R-value to determine the extent of resemblance between the 2-bit defect map representation comprising the scanned defect image and the 2-bit defect map representation comprising the just-stored defect image.

11. A progressive self-learning (PSL) method for enhancing water detect review and classification comprising the steps of

a) accessing a scanned defect image file using an image processing system;

b) determining if the scanned defect image file is empty;

c) ending the PSL classification method when the scanned defect image file is empty;

d) extracting a scanned defect image from the scanned defect image file when the defect image file is not empty;

e) digitizing the scanned defect image;

f) accessing a just-stored digitized defect image database;

g) determining if the just-stored digitized defect image database is empty;

h) manually classifying the scanned defect image when the just-stored digitized defect image database is empty;

i) storing the manually classified scanned defect image as a new image defect type file within the just-stored digitized defect image database when the just-stored digitized defect image database is empty;

j) determining if the scanned defect image aligns with a just-stored digitized defect image disposed within the just stored defect image database when the just-stored digitized defect image database is not empty;

k) accessing the just-stored digitized defect image database when the scanned defect image does not align with a just stored image disposed within the just-stored digitized defect image database

l ) repeating steps f)–)k until the scanned defect image aligns with a just-stored defect image;

m) determining if the scanned defect image matches with a just-stored digitized defect image disposed within the just stored defect image database when the just-stored digitized defect image aligns with the scanned image;

n) accessing the just-stored digitized defect image database when the scanned defect image does not match with the just stored image disposed within the just-stored digitized defect image database;

o) repeating steps f)–n) until the scanned defect image matches with a just-stored defect image; and

p) performing a selfacting classification when the scanned image matches a just-stored defect image accessed from the just-stored digitized defect image database.

12. A progressive self-learning (PSL) method for enhancing wafer or mask defect review and classification comprising the steps of:

performing image processing on a scanned defect image retrieved a scanned defect image file said image processing comprising;

extracting the scanned defect image from a scanned defect image file;

digitizing the scanned defect image using pre-selected digitizing parameters;

transforming the 2-bit defect map into a segmented image according to a square tessellation method; and,

converting the extracted image into a 2-bit defect map comprising two basic shapes including a bulky square region and a round region;

aligning the scanned defect image with a just-stored digitized defect image;

matching the scanned defect image with a just-stored digitized defect image; and

classifying the scanned defect image according to a classification method selected from the group of self-acting and manual.

13. The method of claim 12 wherein the bulky square region comprises a reference to align the scanned defect image with a just-stored digitized defect image.

14. The method of claim 12 wherein the round region comprises a reference in the alignment step to locate a defect position.

15. The method of claim 12 wherein overlapping regions of the scanned defect image and the just-stored digitized defect image are aligned n accordance with an alignment formula.

16. The method of claim 12 wherein the step of matching the scanned detect image with the just-stored digitized defect image comprises the step of:

calculating an R-value to determine the extent of resemblance between the scanned defect image and the just-stored defect image.