Locating shapes in two-dimensional space curves
First Claim
1. A method for finding a feature of an object by finding at least one instance of a model two-dimensional space curve that represents the feature within a target two-dimensional space curve in an image that represents at least a portion of the object, the method comprising the steps of:
- at train-time;
acquiring said model two-dimensional space curve, said model two-dimensional space curve including a plurality of points, each point being characterized by a position in two dimensions, said position being a function of distance S along said model two-dimensional space curve;
for each of said plurality of points of said model two-dimensional space curve, converting said position in two dimensions as a function of distance S into an angle as a function of distance S to provide an angle-based model space curve; and
at run-time;
acquiring said target two-dimensional space curve, said target two-dimensional space curve including a plurality of points, each point being characterized by a position in two dimensions, said position being a function of distance S along said target two-dimensional space curve;
for each of said plurality of points of said target two-dimensional space curve, converting said position in two dimensions as a function of distance S into an angle as a function of distance S to provide an angle-based target space curve; and
using said angle-based model space curve to find at least one position along said angle-based target space curve that results in a match metric value that exceeds a match threshold.
1 Assignment
0 Petitions
Accused Products
Abstract
A method is provided for finding a feature of an object by finding at least one instance of a model two-dimensional space curve that represents the feature within a target two-dimensional space curve that represents at least a portion of the object. Unlike previous methods for finding a two-dimensional space curve within another two-dimensional space curve, the method of the invention is able to find any shape at any orientation or position in the two-dimensional space curve. The model and target two-dimensional space curves each include a plurality of points. Each point is characterized by a position in two dimensions, where the position is a function of distance S along the two-dimensional space curve. For each of the plurality of points of the two-dimensional space curve, the position in two dimensions as a function of distance S is converted into an angle as a function of distance S to provide an angle-based space curve. Then, at run-time, the angle-based model space curve is used to find at least one position along the angle-based target space curve that results in a match metric value that exceeds a match threshold. To provide invariance with respect to rotation, translation, and scaling, it is preferred to use normalized correlation as the match metric. The method of the invention can be used to find the notch or flat on the perimeter of a semiconductor wafer.
-
Citations
17 Claims
-
1. A method for finding a feature of an object by finding at least one instance of a model two-dimensional space curve that represents the feature within a target two-dimensional space curve in an image that represents at least a portion of the object, the method comprising the steps of:
-
at train-time; acquiring said model two-dimensional space curve, said model two-dimensional space curve including a plurality of points, each point being characterized by a position in two dimensions, said position being a function of distance S along said model two-dimensional space curve; for each of said plurality of points of said model two-dimensional space curve, converting said position in two dimensions as a function of distance S into an angle as a function of distance S to provide an angle-based model space curve; and at run-time; acquiring said target two-dimensional space curve, said target two-dimensional space curve including a plurality of points, each point being characterized by a position in two dimensions, said position being a function of distance S along said target two-dimensional space curve; for each of said plurality of points of said target two-dimensional space curve, converting said position in two dimensions as a function of distance S into an angle as a function of distance S to provide an angle-based target space curve; and using said angle-based model space curve to find at least one position along said angle-based target space curve that results in a match metric value that exceeds a match threshold. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
Specification