Method and apparatus for location and inspecting a two-dimensional image including co-linear features
First Claim
1. A method for locating an object in physical space and manipulating said object based on an image of said object acquired in accordance with image space, said method comprising the steps of:
- preprocessing a model of said object by defining co-linear features of said model in accordance with model space, and identifying at least one model set of co-linear feature information;
implementing an image feature finder to find a plurality of features, in image space in a captured image of said object, said plurality of features being substantially similar in character to co-linear features defined in said model during preprocessing;
configuring a co-linear feature finder to receive from said image feature finder point information related to said plurality of features in said captured image and to test said plurality of features for co-linearity and frequency to form a hypothesized line comprised of a first plurality of features;
testing remaining image point information, other than said point information related to said plurality of features forming said hypothesized line, for membership in said hypothesized line until full image line feature information is established;
processing said full image line feature information and said at least one model set of co-linear feature information to generate at least one two dimensional transform mapping said full image line feature information in image space to said at least one model set of co-linear feature information in model space and providing a direct mapping of image space to model space;
using said at least one two dimensional transform to form a hypothesized image including said full image line feature information and projected feature information solved with said direct mapping of image space to model space; and
verifying said hypothesized image against said captured image to determine if each feature is in said captured image and in its proper location.
0 Assignments
0 Petitions
Accused Products
Abstract
A method and apparatus optimizes the process of locating an image of a part by exploiting at least one set of regular co-linear features in the image of the part. A selected one of the at least one set of regular co-linear features is used to quickly correlate a set of image points and a set of model points which are likely to correspond to each other. A one to one correspondence of a subset of model points to a subset of image points yields a single possible location for the part in the image, which location can be quickly and reliably checked for correctness, and if incorrect, a next possible set to set correspondence can be tested until the correct location is found.
59 Citations
25 Claims
-
1. A method for locating an object in physical space and manipulating said object based on an image of said object acquired in accordance with image space, said method comprising the steps of:
-
preprocessing a model of said object by defining co-linear features of said model in accordance with model space, and identifying at least one model set of co-linear feature information; implementing an image feature finder to find a plurality of features, in image space in a captured image of said object, said plurality of features being substantially similar in character to co-linear features defined in said model during preprocessing; configuring a co-linear feature finder to receive from said image feature finder point information related to said plurality of features in said captured image and to test said plurality of features for co-linearity and frequency to form a hypothesized line comprised of a first plurality of features; testing remaining image point information, other than said point information related to said plurality of features forming said hypothesized line, for membership in said hypothesized line until full image line feature information is established; processing said full image line feature information and said at least one model set of co-linear feature information to generate at least one two dimensional transform mapping said full image line feature information in image space to said at least one model set of co-linear feature information in model space and providing a direct mapping of image space to model space; using said at least one two dimensional transform to form a hypothesized image including said full image line feature information and projected feature information solved with said direct mapping of image space to model space; and verifying said hypothesized image against said captured image to determine if each feature is in said captured image and in its proper location. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
-
18. An apparatus for inspecting and locating a two dimensional representation of an object within a captured image, comprising:
-
an image feature finder configured to find a plurality of features, in image space in said captured image of said object, each of said plurality of features being substantially similar in character to features in at least one model set of co-linear features present in a model of said object in accordance with model space; a co-linear feature finder configured to receive from said image feature finder point information related to said plurality of features in said captured image and to test said plurality of features for co-linearity and frequency, said co-linear feature finder configured to form a hypothesized line comprised of a first plurality of features and to identify all features belonging to said hypothesized line in order to obtain full image line feature information; a position hypothesis generator configured to receive said full image line feature information from said co-linear feature finder and said at least one model set of co-linear feature information, and to generate at least one two dimensional transform mapping said full image line feature information in image space to said at least one model set of co-linear feature information in model space effecting a direct mapping of image space to model space; a verification processor configured to receive said at least one two dimensional transform from said position hypothesis generator and to form a hypothesized image including said full image line feature information and projected feature information solved with said direct mapping of image space to model space, said verification processor being further configured to verify said hypothesized image against said captured image to determine if each feature is in said captured image and in its proper location. - View Dependent Claims (19, 20, 21, 22, 23, 24, 25)
-
Specification