Determining the Uniqueness of a Model for Machine Vision

• US 20120170835A1
• Filed: 12/29/2010
• Published: 07/05/2012
• Est. Priority Date: 12/29/2010
• Status: Active Grant

• Current Assignee: Cognex Corporation
• Original Assignee: Cognex Corporation
• Inventors: Wang, Xiaoguang, Jacobson, Lowell
• Granted Patent: US 8,542,912 B2
• Time in Patent Office: Days
• US Class Current: 382/159
• CPC Class Codes: G06F 18/2193 based on specific statistical tests
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries

Abstract

Described are methods and apparatuses, including computer program products, for determining model uniqueness with a quality metric of a model of an object in a machine vision application. Determining uniqueness involves receiving a training image and a first set of model parameters, generating a first model of an object, generating a second model of the object based on the training image and a second set of model parameters modified from the first set of model parameters, determining a set of poses that represent possible instances of the second model in the training image, and computing a quality metric of the first model based on an evaluation of the set of poses with respect to the training image.

Claims

1. A computerized method for determining a quality metric of a model of an object in a machine vision application, the method comprising:

receiving a training image and a first set of model parameters;

generating a first model of an object;

generating a second model of the object based on the training image and a second set of model parameters modified from the first set of model parameters;

determining a set of poses that represent possible instances of the second model in the training image; and

computing a quality metric of the first model based on an evaluation of the set of poses with respect to the training image.

2. The method of claim 1 wherein modifying the first set of model parameters to produce the second set of model parameters comprises perturbing one or more values in the first set of model parameters.

3. The method of claim 1 wherein the model comprises a geometric description of the object in the training image.

4. The method of claim 1 wherein the model comprises a portion of the training image.

5. The method of claim 1 further comprising computing at least a primary score and a secondary score for at least a portion of the set of poses.

6. The method of claim 5 wherein computing the quality metric of the model is based on the primary score and the secondary score.

7. The method of claim 5 wherein computing the quality metric of the model is based on a distribution of the computed scores for the portion of the set of poses.

8. The method of claim 1 further comprising:

generating a plurality of models based on the training image and a plurality of different sets of model parameters, the plurality of different sets of model parameters being based on modifications to the first set of model parameters; and

determining a set of poses for each of the plurality of models, each set of poses representing possible instances of one of the plurality of models in the training image;

wherein computing the quality metric is further based on an evaluation of the sets of poses, for each of the plurality of models, with respect to the training image.

9. The method of claim 8 further comprising computing at least primary scores and secondary scores for at least a portion of the first set of poses and at least portions of the sets of poses computed for each of the plurality of models.

10. The method of claim 9 wherein determining the quality metric of the model is based on the secondary scores.

11. The method of claim 9 wherein determining the quality metric of the model is based on a distribution of the computed scores for the portion of the first set of poses and distributions of the computed scores for the portions of the sets of poses computed for each of the plurality of models.

12. The method of claim 1 further comprises modifying the training image.

13. The method of claim 12 wherein modifying the received training image comprises adding noise to the received training image.

14. The method of claim 1 wherein determining the set of poses comprises modifying one or more search parameters.

15. The method of claim 14 wherein the one or more search space parameters comprise a starting pose value, one or more search range values, or any combination thereof.

16. A computer program product, tangibly embodied in a machine-readable storage device, the computer program product including instructions being operable to cause a data processing apparatus to:

receive a training image and a first set of model parameters;

generate a first model of an object;

generate a second model of the object based on the training image and a second set of model parameters modified from the first set of model parameters;

determine a set of poses that represent possible instances of the second model in the training image; and

compute a quality metric of the first model based on an evaluation of the set of poses with respect to the training image.

17. A system for determining a quality metric of a model of an object in a machine vision application, the system comprising:

interface means for receiving a training image and a first set of model parameters;

model generating means for generating a first model of an object;

model generating means for generating a second model of the object based on the training image and a second set of model parameters modified from the first set of model parameters;

processor means for determining a set of poses that represent possible instances of the second model in the training image; and

processor means for computing a quality metric of the first model based on an evaluation of the set of poses with respect to the training image.

18. A machine vision system for determining a quality metric of a model of an object in a machine vision application, the system comprising:

an interface for receiving a training image and a first set of model parameters;

a model generating module For generating a first model of an object and for generating a second model of the object based on the training image and a second set of model parameters modified from the first set of model parameters;

a run-time module for determining a set of poses that represent possible instances of the second model in the training image; and

a quality-metric module for computing a quality metric of the first model based on an evaluation of the set of poses with respect to the training image.