Method and apparatus for analyzing image data generated during underground boring or inspection activities

• US 9,219,886 B2
• Filed: 03/13/2013
• Issued: 12/22/2015
• Est. Priority Date: 12/17/2012
• Status: Active Grant

• Current Assignee: Emerson Electric Company
• Original Assignee: Emerson Electric Company
• Inventors: Sooy, Josh, Daw, Justin, Fresty, Paul, Muresan, Lucian-Vasile, Yuronich, Brad
• Primary Examiner(s): Torrente, Richard
• Assistant Examiner(s): MATT, MARNIE A
• Application Number:
  US13/801,330
• Publication Number:
  US 20140168408A1
• Time in Patent Office: 1,014 Days
• Field of Search: 348/79, 348/85
• US Class Current: 1/1
• CPC Class Codes: E21B 47/002 by visual inspection
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: H04N 7/18 Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Abstract

A system includes a visual inspection system and an image analysis system. The visual inspection system includes an inspection camera that captures images from within an interior of at least one of a utility line and a tunnel for installing a utility line, and a first communication interface that communicates image data corresponding to the images. The image analysis system includes a second communication interface that receives the image data from the visual inspection system, a model adaptation module that modifies a classifier model based on at least one of feedback data and training data, and a classifier module that implements the classifier model to identify a plurality of features in the image data corresponding to defects and that modifies the image data according to the identified plurality of features. The defects include at least one of a cross-bore, a lateral pipe, and an imperfection.

Claims

1. A system for determining whether a newly installed utility line or a tunnel for installing a utility line bores through an existing utility line, the system comprising:

a visual inspection system, including

an inspection camera that captures images from within an interior of at least one of the existing utility line and the tunnel for installing the utility line, and

a first communication interface that communicates image data corresponding to the images; and

an image analysis system including

a second communication interface that receives the image data from the visual inspection system,

a model adaptation module that modifies a classifier model based on at least one of feedback data and training data, and

a classifier module that

(i) implements the classifier model to identify a plurality of features in the image data corresponding to defects, wherein the defects include at least one of a cross-bore and a lateral pipe, wherein the classifier model categorizes each of the plurality of the features, wherein each of the plurality of features is indicative of at least one of the defects, wherein at least one of the plurality of features corresponding to the cross-bore indicates, from within the tunnel for installing the utility line, that the tunnel bores through the existing utility line, and wherein at least one of the plurality of features corresponding to the lateral pipe indicates, from within the existing utility line, that the newly installed utility line passes through the existing utility line, and

(ii) modifies the image data according to the identified plurality of features.

2. The system of claim 1 wherein the image analysis system is at least one of remotely located from the visual inspection system and integrated within the visual inspection system.

3. The system of claim 1, wherein the image analysis system receives the image data from the visual inspection system via a cloud networking system.

4. The system of claim 3, wherein one or more functions of the image analysis system are implemented using the cloud networking system.

5. The system of claim 1, wherein the classifier model assigns a probability to each of the plurality of features, and wherein each of the probabilities corresponds to a probability that a respective one of the defects is present in the image data.

6. The system of claim 1, wherein modifying the image data according to the identified features includes assigning a label to a frame of the image data that includes one of the identified features.

7. The system of claim 6, wherein the label includes an indication of at least one of a lateral pipe, no lateral pipe, a crossbore, and no crossbore in the frame.

8. The system of claim 1, wherein the training data includes training image data including the plurality of features.

9. The system of claim 8, wherein the model adaptation module modifies the classifier model based on whether the classifier model identifies the plurality of the features in the training image data.

10. The system of claim 1, wherein the plurality of features include at least one of parallel lines, color information, K-means clustering, and gradients.

11. The system of claim 1, wherein the model adaptation module extracts the plurality of features from the training data, labels the image data according to the plurality of features extracted from the training data using the classifier model, stores classified training data corresponding to the labeled image data, compares the classified training data to test data indicative of actual features included in the training data, and updates the classifier model according to the comparison.

12. The system of claim 1, wherein the feedback data corresponds to feedback provided by a user indicative of whether the plurality of features identified by the classifier module are accurate.

13. The system of claim 1, wherein modifying the image data includes indicating a soil type indicated by the plurality of features.

14. The system of claim 1, wherein the classifier model assigns a probability to one of the identified features that at least one of the defects is present in the image data based on a number of times a user indicated that the one of the identified features corresponded to the at least one of the defects.

15. The system of claim 1, wherein identifying the plurality of features in the image data corresponding to defects includes using at least one of parallel lines, K-means clustering, and gradients to identify the at least one of the cross-bore and the lateral pipe.

16. A method for determining whether a newly installed utility line or a tunnel for installing a utility line bores through an existing utility line, the method comprising:

using an image inspection system, capturing images from within an interior of at least one of the existing utility line and the tunnel for installing the utility line; and

using an image analysis system,

receiving image data corresponding to the images;

modifying a classifier model based on at least one of feedback data and training data;

using the classifier model to (i) identify a plurality of features in the image data corresponding to defects and (ii) categorize each of the plurality of the features, wherein each of the plurality of features is indicative of at least one of the defects, wherein the defects include at least one of a cross-bore and a lateral pipe, wherein at least one of the features corresponding to the cross-bore indicates, from within the tunnel for installing the utility line, that the tunnel bores through the existing utility line, and wherein at least one of the features corresponding to the lateral pipe indicates, from within the existing utility line, that the newly installed utility line passes through the existing utility line; and

modifying the image data according to the identified plurality of features.

17. The method of claim 16 wherein the image analysis system is at least one of remotely located from the visual inspection system and integrated within the visual inspection system.

18. The method of claim 16, wherein the image analysis system receives the image data via a cloud networking system.

19. The method of claim 16, further comprising assigning, using the classifier model, a probability to each of the plurality of features, wherein each of the probabilities corresponds to a probability that a respective one of the defects is present in the image data.

20. The method of claim 16, wherein modifying the image data according to the identified features includes assigning a label to a frame of the image data that includes one of the identified features.

21. The method of claim 20, wherein the label includes an indication of at least one of a lateral pipe, no lateral pipe, a crossbore, and no crossbore in the frame.

22. The method of claim 16, wherein the training data includes training image data including the plurality of features.

23. The method of claim 22, further comprising modifying the classifier model based on whether the classifier model identifies the plurality of the features in the training image data.

24. The method of claim 16, wherein the plurality of features include at least one of parallel lines, color information, K-means clustering, and gradients.