High-speed railroad inspection using coordinated 3D cameras

• US 9,049,433 B1
• Filed: 01/06/2012
• Issued: 06/02/2015
• Est. Priority Date: 01/06/2012
• Status: Expired due to Fees

• Current Assignee: John H. Prince
• Original Assignee: John H. Prince
• Inventors: Prince, John H.
• Primary Examiner(s): Topgyal, Gelek W
• Application Number:
  US13/374,678
• Time in Patent Office: 1,243 Days
• Field of Search: 348/47, 348/148, 382/104, 382/154
• US Class Current: 1/1
• CPC Class Codes: B61K 9/10 for detecting cracks in rails or welds thereof
• CPC Class Codes: B61L 23/042 Track changes detection
• CPC Class Codes: B61L 23/044 Broken rails
• CPC Class Codes: B61L 23/045 Rail wear
• CPC Class Codes: B61L 27/53 for trackside elements or systems, e.g. trackside supervision of trackside control system conditions
• CPC Class Codes: G01N 2021/1785 Three dimensional
• CPC Class Codes: G01N 2021/8887 based on image processing techniques
• CPC Class Codes: G01N 2021/9518 using a surface follower, e.g. robot
• CPC Class Codes: G01N 21/9515 Objects of complex shape, e.g. examined with use of a surface follower device measuring contours and curvatures G01B11/24
• CPC Class Codes: G01N 2291/2623 Rails; Railroads
• CPC Class Codes: G06T 2207/10021 Stereoscopic video; Stereoscopic image sequence
• CPC Class Codes: G06T 2207/30108 Industrial image inspection
• CPC Class Codes: G06T 7/0004 Industrial image inspection
• CPC Class Codes: G06T 7/001 using an image reference approach
• CPC Class Codes: H04N 13/239 using two 2D image sensors having a relative position equal to or related to the interocular distance H04N13/243 takes precedence
• CPC Class Codes: H04N 13/243 using three or more 2D image sensors

Abstract

To emulate the 3D visual acuity of an individual track-walker, matched pairs of cameras (which simulate our human eyes) are mounted on a moving vehicle above the tracks. The cameras are mounted both transversely and longitudinally, to create 3D images in two orthogonal orientations, which helps to eliminate visual voids. The location and orientation of each pair of cameras is determined by its specific task, in the present case: (i) to examine each left or right rail for anomalies (two pairs), and (ii) to examine every tie for damage (at least one pair). The camera pairs enable measurements such as defect depth and track width variations. The images are compressed in real time for local storage or high-speed transmission for remote display. The basic arrangement of cameras can be amplified or modified as necessary to circumstances.

Claims

1. A method of detecting worn and defective surfaces on objects needing optical inspection comprising the steps of:

moving an array of imaging devices relative to said objects over a pathway, at least one pair of said imaging devices positioned as a first and second imaging device and having an included angle about a reference center plane to generate corresponding uncorrected first and second images of said objects;

capturing a sequence of image data formed as said imaging devices move over said pathway;

forming a sequence of corrected 3D images from said uncorrected first and second images by performing an alignment of said first and second images mechanically and/or electronically;

for each said pair of imaging devices, comparing at least one of said first or said second corrected 3D images against a reference image to identify a defective or worn surface of said object needing inspection;

where said worn and defective surface is on a railroad track component including at least one of;

a rail, a rail edge, nuts, tie-plates, base-plates, pandrol clips, or ties.

2. The method as in claim 1 where said comparing at least one of said first or said second corrected 3D image against a reference image is a comparison between said first and said second corrected 3D images.

3. The method as in claim 1 wherein said alignment includes the steps of:

coarsely aligning said imaging devices by means of physical manipulation; and

finely aligning said imaging devices by means of image matching software examining features in said images to achieve said alignment.

4. The method as in claim 1 where said comparing said first corrected 3D image and said second corrected 3D image is a comparison between said first corrected 3D image and a corresponding reference first image, and/or a comparison between said second corrected 3D image and a reference second image.

5. The method as in claim 4 wherein said imaging devices detect a surface defect such as a crack or erosion by detecting the offset of an imaged feature in at least one of said pair of images.

6. The method as in claim 5 wherein the depth of said defect is proportional to said offset in said at least one of said pairs of images.

7. The method of claim 1 where said comparison includes a step of storing said first and second corrected 3D image as a stereo pair.

8. The method of claim 7 where said step of storing includes location information associated with said defective or worn surface.

9. The method of claim 8 where said location information is a GPS coordinate acquired at the time said uncorrected 3D images were acquired.

10. The method of claim 1 where each said first and second pair of imaging devices is oriented to intersect a common point of a potentially worn or defective surface to be inspected.

11. The method of claim 1 where said corrected 3D images are generated by aligning said first and second image by minimizing ESSD(u) where:

ESSD(u)=Σ

i[I1(xi+u)−

I0(xi)]²=Σ

i(ei)²

and u=(u, v) is a feature displacement on orthogonal axes

and ei=I1 (xi+u)−

I0(xi) is the error function or feature displacement offset within feature areas.

12. The method of claim 1 where said worn and defective surface is at least one of cracks, erosion, brinelling, or spalling in a rail.

13. An imaging device comprising:

a plurality of imaging detectors arranged in pairs and mounted on a structure to allow viewing of said fixed objects from different perspectives;

said pairs of detectors aligned in pairs or in a plurality for 3D viewing of said fixed objects, each said pair of detectors separated by a distance sufficient to resolve the depth of a defect by a variance in an imaged feature from a reference dimension;

said rigid structure providing mounting for at least one pair of detectors aligned parallel to a structure to be imaged and also at least one pair of detectors arranged substantially perpendicular to said one pair of image detectors;

wherein said certain fixed objects are railroad ties and other fixed objects include at least one of;

rails, baseplates, clips, ties, or other objects associated with the tracks of a railroad.

14. The imaging device as in claim 13 where at least one said pair of detectors compare a corresponding pair of images from said pair of detectors to identify a defect.

15. The imaging device as in claim 13 where said pair of detectors compares acquired images to reference images.

16. The imaging device as in claim 13 where at least one said pair of detectors detect a surface defect such as a crack or erosion by detecting the offset of an imaged feature in at least one of a pair of images acquired by said pair of detectors.

17. The imaging device as in claim 16 where the depth of a detected defect is proportional to said offset in said at least one of said detectors.

18. The imaging device as in claim 13 wherein said imaging means and processing means calculate defect metrics including depth, erosion, and track width as said plurality of image detectors is moved over said objects.