SONAR IMAGING

• US 20140079287A1
• Filed: 09/18/2013
• Published: 03/20/2014
• Est. Priority Date: 09/18/2012
• Status: Active Grant

• Current Assignee: SeeByte Ltd. (General Dynamics Corporation)
• Original Assignee: SeeByte Ltd. (General Dynamics Corporation)
• Inventors: Daniell, Oliver
• Granted Patent: US 9,235,896 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G01S 15/89 for mapping or imaging
• CPC Class Codes: G01S 7/539 using analysis of echo signal for target characterisation; Target signature; Target cross-section
• CPC Class Codes: G06T 7/74 involving reference images or patches

Abstract

A method for recognising a target in a sonar image, the method comprising: normalising a sonar image; using/defining multiple test objects; rotating each test object between multiple positions; using a projection of each test object in each position as a template, so that multiple templates are provided for each test object, each template corresponding to a different rotational position; applying the multiple templates for the multiple test objects to the normalised image; and creating at least one feature vector for the image for use in target recognition.

Claims

1. A method for recognising a target in a sonar image, the method comprising the steps of:

normalising a sonar image;

at least one of using or defining multiple test objects;

rotating each test object between multiple positions;

using a projection of each test object in each position as a template, so that multiple templates are provided for each test object, each template corresponding to a different rotational position;

applying the multiple templates for the multiple test objects to the normalised image; and

creating at least one feature vector for the image for use in target recognition.

2. A method as claimed in claim 1 wherein each template comprises at least a highlight region and a shadow region.

3. A method as claimed in claim 2, wherein a plurality of classes of template is defined for each template.

4. A method as claimed in claim 3, wherein the plurality of classes of template comprise a standard template having a highlight and a shadow region only.

5. A method as claimed in claim 3, wherein the plurality of classes of template comprise a bounded front template, in which a region is added before the highlight region in an across-track direction, for example a region the same size as the highlight region is added before the highlight region in an across-track direction.

6. A method as claimed in claim 3, wherein the plurality of classes of template comprise a bounded side template, in which each side of the combined highlight and shadow region is bounded.

7. A method as claimed in claim 1, wherein applying the multiple templates for the multiple test objects to the normalised image to detect an object in the normalised image involves convolving the multiple templates with the normalised image.

8. A method as claimed in claim 1, further comprising the steps of:

using the normalised image to define a plurality of skewed integral images, each for a different skew angle; and

applying the multiple templates for the multiple test objects to the skewed integral images.

9. A method as claimed in claim 8, wherein the skewed integral images are skewed at angles that correspond to the different rotational positions of the templates.

10. A method as claimed in claim 1, wherein the step of normalising features in the sonar image involves normalising the whole image and then applying the multiple templates or normalising one pixel of the image based on local image statistics and then applying the multiple templates for that pixel.

11. A method as claimed in claim 1, further comprising the step of determining at least one feature value for each template, the feature value for each template being included in the feature vector.

12. A method as claimed in claim 11, wherein when a plurality of classes of template are provided the method further involves determining a feature value for each template class and adding the feature value for each template class to the feature vector.

13. A method as claimed in claim 12, further comprising the step of using the feature vector for target recognition.

14. A method as claimed in claim 12, further comprising the step of using the feature vector in a target recognition classifier.

15. A method as claimed in claim 14 wherein the target recognition classifier comprises a supervised machine learning algorithm.

16. A method as claimed in claim 1, further comprising the step of capturing the sonar image.

17. A computer implemented target detection system, said system comprising one or more memory storage areas and one or more computer processors configured to perform the method of claim 1.

18. A non-transitory computer program product comprising at least one computer-readable storage medium having computer-readable program code portions embodied therein, the computer-readable program code portions comprising one or more executable portions configured for performing the method of claim 1.