Method for pattern inspection
AssigneeTokyo Seimitsu (Israel) Ltd.
InventorsMark Geshel, Niv Shmueli, Gideon Friedmann, Orna Bregman-Amitai
Life of this patent
Estimating the term…
Abstract
A method for inspecting a patterned surface employs reference data related to the pattern to provide a map for identifying regions which are expected to generate equivalent images. These regions are then compared in an image-to-image comparison to identify possible defects. In a first implementation, the regions are related by local symmetry operators. In a second, disjoint corner features or other features are classified and similar features compared.
Claims27
A method for inspecting a patterned surface having a pattern which corresponds substantially to a set of reference data, the method comprising:
- (a) generating a multilevel pixel image of at least one region of the pattern, the pattern being non-repetitive within said at least one region;
- (b) processing the reference data corresponding to the at least one region of the pattern to identify at least two groups of pixels which are expected to generate equivalent images;
- and
- (c) comparing the at least two groups of pixels from the multilevel pixel image.
A method for inspecting a patterned surface having a pattern which corresponds substantially to a set of reference data, the method comprising:
- (a) generating a multilevel pixel image of at least one region of the pattern;
- (b) processing the reference data corresponding to the at least one region of the pattern to identify at least one group of pixels which is expected to be invariant under a given pixel displacement vector;
- and
- (c) comparing the at least one group of pixels from the multilevel pixel image with pixels related thereto by said pixel displacement vector.
A method for inspecting a patterned surface having a pattern which corresponds substantially to a set of reference data, the method comprising:
- (a) processing the reference data to identify regions of the pattern which include defined features;
- (b) classifying said regions by use of the reference data into sets of regions containing comparable occurrences of said defined features which are expected to be similar according to predefined criteria;
- (c) obtaining multilevel pixel images of a plurality of said regions which contain said comparable occurrences;
- and
- (d) comparing said multilevel pixel images for said comparable occurrences.
Dependencies are inferred from the claim text.