GB2524006A

GB2524006A - Signage testing

Info

Publication number: GB2524006A
Application number: GB1404182.6A
Authority: GB
Inventors: Craig Phillips
Original assignee: Individual
Current assignee: Individual
Priority date: 2014-03-10
Filing date: 2014-03-10
Publication date: 2015-09-16
Also published as: GB201404738D0; GB201404182D0

Abstract

The invention concerns a signal alignment test performed by a processing unit 6 that can be used to automatically check correct configuration of signage 8 on a stretch of railway track or road. The method comprises: receiving image or video data from a section of road or track; identifying from the data, shapes as target signs for which to calculate attributes (e.g. alignment, orientation); generating output data about the set-up of the target signs. The method could include taking data from a camera mounted on the vehicle or locomotive 4. The processing unit contains software or firmware to calculate deviations in signage data such as face angle, largest perceived surface area, and effective luminosity, for each sign viewable on the given stretch of track or road. Claims are also included for a system having a memory that stores reference image or video data indicative of an acceptable condition of the signage. By comparing images recorded by the camera to the stored data the system can issue an alert if the signage is in an unacceptable condition.

Description

SIGNAGE TESTING

The invention relates to a system arid method for checking signage, for example rail signage, which may include, but is not limited to, checking the alignment and/or effective luminosity of signage, e.g. signals, laminated signs, LED signs, displaying warnings, speed limits, and other infbrmation.

The invention could also be used in a road environment.

In a rail environment, it is important that signs are aligned correctly so the driver can adequately see them with enough notice to act on the information conveyed, Similarly, they must meet luminosity requirements.

In an underground network or a tunnel, brake dust and other pollutants can build up on the face of signs, impeding luminosity. In overground networks, other factors such as foliage can obscure part or all of a sign, impeding visual interpretation.

When signage is designed and installed at a railway site, it is careftilly setup within defined tolerances, so that the driver can see the sign clearly from his approach. Over time, vibration and other factors can cause the original setup ofthe sign to change. If the sign's face has altered in angle since installation, it may no longer be effective in alerting the driver in good time.

Routine checking of signage is carried out by rail network operators, Additionally, spot checks will be carried out after incidents to check signage is correct. The incumbent method is to send a locomotive down the track at slow speed outside of operating hours, with a party of experts to assess the signage.

This incumbent method is not only expensive, it prevents other maintenance activities for the duration of the assessment. Furthermore, it is a subjective method of assessment.

The invention according to a first aspect provides a system for peiforming an alignment test on signage, comprising: a processor arranged to receive image or video data of signage from a camera mounted on a locomotive or vehicle arid to perform shape recognition on the image or video data so as to recognise shapes as target signs on S which to calculate attributes, and, using calculated attributes to generate output data indicative of the set-up of target signs within a given track or road section.

The processor may be arranged such that, for a given sign, angles are calculated from the shape outlines and used to indicate a measure of alignment of the given sign.

The processor may be arranged such that, for a given sign, its shape's outline measurements are used to indicate a measure of alignment of the given sign.

The processor may be arranged such that the outline measurements are expressed as a ratio to represent a measure of sign alignment.

The processor may be arranged such that thee angles are calculated from outline measurements, and used as a measure of sign alignment.

The processor may be arranged such that, for a given sign, calculation of its shape's largest surface area is used to predict a repeatable position on a section of track or road in which to record the given sign's attributes or used as an indication ofmisalignment of the given sign's face.

The processor may be arranged such that, for a given sign, the variance of brightness of a set of pixels enclosed within the shape's outline is used as an indication of the sign's effective luminosity.

The system may further comprise memory on which is stored a datum or norm report for a section of track or road, on which signage has been confirmed to be set-up correctly to current standards by traditional surveying methods, and wherein the output data generated by the processor is based on a comparison of the incoming video or image data.

A second aspect of the invention provides a system for mounting on a vehicle or S locomotive, the system comprising a memory storing a set of im age or video data and/or reference data indicative of an acceptable condition of signage in terms of alignment and/or luminosity for a given length of road or track, and a processor configured to receive an image or video data feed from a camera mounted on said vehicle or locomotive, to compare said received image or video data with the reference data, and to automatically issue an alert or report in the event that either or both of the alignment or luminosity for signage is outside of a predetermined tolerance.

A third aspect of the invention provides a method for performing an alignment test on signage, e.g. on a vehicle or locomotive, the method comprising: receiving image or I S video data from a section of road or track; identifying from the data shapes as target signs on which to calculate attributes; and generating output data on the set-up of target signs within the section of road or track.

The method may be performed by a computer program, e.g. one stored on a non-transitory memory, when executed by a processor.

A preferred embodiment of the invention will now be described with reference to the accompanying drawings in which: FIGURE lisa perspective view of a section of railway track, with a target signal and a locomotive employing a system which is an embodiment of the present invention; FIGURE 2 shows the flow diagram of processing steps performed by an embodiment of the invention; FIGURE 3 shows target sign close-ups, as captured or seen by the Figure 2 embodiment; FIGURE 4 shows one method for deternting a measure offàce angle for a sign; and FIGURES shows one method for determining effective luminosity of a sign.

A preferred embodiment employs a processing unit that takes a data feed from a camera system mounted on a locomotive. The processing unit typically comprises one or more processors or controllers operating under program control, with the processing steps outlined below being part of a suitable computer program. Firmware could also be S provided. The system and method provides an automatic, repeatable and objective method ofexaminirg signs, in this case in a rail environment. The system and method provides a way of determining a number of sign attributes, including a measure of face angle, using shape recognition thnctions/algorithms.

Referring to Figure 1, a high definition camera 2 is placed on the front exterior of a locomotive 4, as close to the driver's optimal viewing position as feasible, pointing in the direction of travel. The camera data feed is provided to a processing unit 6 which is the present embodiment. In typical operation, this locomotive 4 continues a route as it normally would, with driver and passengers. A typical sign 8 is indicated along one side I S of the track.

The camera 2 is used to receive and / or record images, The processing unit 6 creates a dataset containing various data for each sign it records.

An optimal measuring point 10 is indicated. This data contains, but is not necessarily limited to, largest perceived surface sign area (A), angle of horizontal deviation (0), angle of vertical deviation (4)) and effective luminosity (L).

In an example use of this embodiment, a given route would be surveyed manually to confirm that every sign is setup correctly. Then, the route would be driven by the locomotive 4 canying the processing unit 6 to create a datum' dataset for that route.

This could be done post-maintenance, or post installation of new signage, Subsequent test runs on that route would create a measured dataset, which would be compared against the datum dataset to check tolerances of A, 0, 4), L, and raise any breaches, During a test run, the processing unit 6 is arranged to perform several processing steps, shown in Figure 2 and outlined below, Step 2. 1: the camera feed is received for processing, either in real-time, or after a route is completed.

Step 2,2: shape recognition sothvare identifies the outlines of signs, be they square, circular, elliptical, or rectangular, using edge-detection techniques commonplace in the practise of image analysis.

Step 2,3: an optimal measuring point is identified 10, Note that there exists a section of track on which the locomotive travels, where the sign must be clearly visible, as specified in safety standards. See Figure 1 for illustration. It is assumed that the point at which the largest surface area of the sign S is visible from the driver's POV (point of view) lies within this section of track, Therefbre, a measurement for a target sign 8 can be taken at the point in time where the area enclosed by the sign's outline is at its largest from the camera POV. This area is denoted by A. Another way to calculate a measuring position would be to use a location feed, such as from a positioning system, e.g. GPS, and check this against a dataset with preset optimal positions. Clearly, the former method is simpler, but maintains the advantage of being repeatable, so in many cases is preferable.

For every sign on a given route, at the measuring point, a still image is extracted from the video feed. This still image contains information on the shape's coordinates determined by shape recognition software, Step 2.4: For each still image, the processing software determines a measure of the sign's thee alignment. This can be expressed as a ratio of the outline measurements.

Since misalignment causes distortion of a shape as seen by the camera (see Figure 3), this ratio contains deviation information.

This ratio can be further expressed as a sign's perceived thee angle as a function of the S shape's distortion, as shown in Figure 4. In this proposed implementation, deviation angles are calculated in two planes, yielding (E) and if).

Additionally, for square or rectangular signs, an angle representing deviation in the z-axis F could be calculated, by measuring the angle between the top edge of the shape outline and a bisecting horizontal axis across the still image.

Additionally, an effective luminosity reading for the sign can be taken from the still image by examining the brightness levels of all pixels enclosed by the shape's outline (within surface area of shape), In this proposed implementation, L = variance of pixel brightness of entire shape. See Figure 5.

Step 2.5: Measured L can be compared against the datum L for that sign, and if measured L is lower (by a given tolerance, e.g., a percentage), it can be concluded that the luminosity of the sign has been rendered ineffective by dust or pollutant build up, and this breach would be raised to the relevant authority for remedial action via means of a report.

If the largest perceived surface area of a sign on the still image, A, is outside tolerance for that sign, this would be reported. Also, if no shape is recognised throughout a section of track, this could also be reported. Also, if an additional shape is found in a section of track where a new sign has been added, this could also be reported.

If an angle 0, C, or F, when compared against its datum, has breached tolerance, this would be reported as a sign misalignment. An alternative method would be to trigger on a breach oftolerance for a orb or ratio of a and b.

So, following the steps in Figure 2, the embodiment described herein fulfils the objective of an automatic signage examination test that can operate simultaneously to the normal service of a railway. In practise, some of the steps of image analysis described may occur within a single processing step.

It will be appreciated that the embodiment can be applied to many modes of transportation including, but not limited to, overground rail, underground rail, tram rail, and highways.

It is desirable to describe in more detail an example method for calculating the face angles e and D of a sign from a set of Cartesian measurements, as in step 2.4 in Figure 2, as this demonstrates the invention's usefi.ilness when applied to signage installation standards expressed angularly.

Referring to Figure 4, if we assume the target sign has a square outline, and we assume that the lens of camera 2 is situated perpendicular to the centre of the sign, and we have the measured pixel lengths (for square pixels) of the rectangle incident on the camera lens, a and b, then the deviation angle 0, can be calculated as the arc cosine of b over a, and the deviation angle), can be calculated as the arc cosine of a over b, as shown in the formulae of Figure 4.

Clearly, it is unlikely that the camera lens will be exactly perpendicular to a correctly-set-up sign; i.e., angles of zero deviation are not expected. What is of concern is the delta between the measured angle and the datum, Similarly, many signs are rectangular rather than square, so even with a perpendicular set-up, the calculations given wouldn't result in zero angles. It is the datum that is of concern, calculated when the sign is proven to be set-up correctly by traditional means of surveying.

In the case of signs with elliptical or circular outlines, the rectangle that encloses the shape would be passed from the image analysis step (step 2.2 of Figure 2) and the calculation of face angle would proceed as described for a rectangle.

It should be noted that simply the ratio of a and b could be used by the embodiment, without recourse to (5) or P, but installation standards are likely specified as tolerances of angles, so information expressed as angular is more usethl to the end-user, However, in another embodiment, the processing unit 6 may record a and b or the ratio of these two measurements and not calculate angles 0 or (I).

It should be fUrther noted that, upon calculation of one angle (0 or I), the other one could be derived, so an example enthodiment may not record both® and (* I.e: If we say the ratioi' = then cosø=-and cose=r so, see 2= r therefore, 9 arccos(sec I') It is possible that a target sign could be out of alignment in both planes, such that the calculated angles are within tolerance to their datums, even though the sign is out of alignment. This is why it is necessary to record the largest perceived surthce area of the sign, A, because in this scenario, A would differ from its datum, and a tolerance breach would be recognised on this basis.

It is further possible, however unlikely, that the alignment in any plane could be out at an angle of 180 degrees, such that the sign appears in-tolerance, even though it has flipped around completely.

It will be appreciated that the above described embodiments are purely illustrative and are not limiting on the scope of the invention. Other variations and modifications will be apparent to persons skilled in the art upon reading the present application.

Moreover, the disclosure of the present application should be understood to include any novel features or any novel combination of features either explicitly or implicitly disclosed herein or any generalization thereof and during the prosecution of the present application or of any application derived therefrom, new claims may be ibrmulated to cover any such features and/or combination of such features.

Claims

Claims 1 A system for performing an alignment test on siiage, comprising: a processor arranged to receive image or video data of signage from a camera mounted on a locomotive or vehicle and to perform shape recognition on the image or video data so as to recognise shapes as target signs on which to calculate attributes and using the calculated attributes to generate output data indicative of the set-up of target signs within a given track or road section.
2. A system according to claim 1, wherein the processor is arranged such that angles are calculated from the shape outlines and used to indicate a measure of alignment of the given sign.
3 A system according to claim t, wherein the processor is arranged such that, for a given sign, its shape's outline measurements are used to indicate a measure of alignment of the given sign.
4, A system according to claim 3, wherein the processor is arranged such that the outline measurements are expressed as a ratio to represent a measure of sign alignment.
5, A system according to any of claims 2 to 4, wherein the processor is arranged such that face angles are calculated from outline measurements, and used as a measure of sign alignment.
6. A system according to any preceding claim, wherein the processor is arranged such that, for a given sign, calculation of its shape's largest surface area is used to predict a repeatable position on a section of track or road in which to record the given sign's attributes or used as an indication ofmisalignment of the given sign's face,
7. A system according to any preceding claim, wherein the processor is arranged such that, for a given sign, the variance of brightness of a set of pixels enclosed within the shape's outline is used as an indication of the sign's effective luminosity.
8. A system according to any preceding claim, thither comprising memory on which is stored a datum or norm report for a section of track or road, on which signage has been confirmed to be set-up correctly to cunent standards by traditional surveying methods, and wherein the output data generated by the processor is based on a S comparison ofthe incoming video or image data.
9, A system for mounting on a vehicle or locomotive, the system comprising a memory storing a set of image or video reference data indicative of an acceptable condition of signage in terms of alignment and/or luminosity for a given length of road or track, and a processor configured to receive an image or video data feed from a camera mounted on said vehicle or locomotive, to compare said received image or video data with the reference data, and automatically to issue an alert or report in the event that either or both of the alignment or luminosity tbr signage is outside of a predetermined tolerance.
10. A method for performing an alignment test on signage, e.g. on a vehicle or locomotive, the method comprising: receiving image or video data from a section of road or track; identifying from the data shapes as target signs on which to calculate attributes; and generating output data on the set-up of target signs within the section of road or track.
11. A method according to claim 10, wherein, for a given sign, its shape's outline measurements are used to indicate a measure of alignment of the given sign.
12. A method according to claim 1 I, wherein outline measurements are expressed as a ratio to represent a measure of signal alignment.
13. A method according to claim 11 or claim 12, wherein face angles are calculated from outline measurements, and used as a measure of sign alignment.
14. A method according to any of claims 10 to 13, wherein, for a given sign, calculation of its shape's largest surface area is used to (i) predict a repeatable position on a section of track in which to record the given sign's attributes, or (ii) used as an indication of misalignment ofthe given sign's face.S15. A method according to any of claims 10 to 14, wherein, for a given sign, the variance of brightness of a set of pixels enclosed within the shape's outline is used as an indication of the sign's effective luminosity.16. A method according to any of claims 10 to 15, wherein a datum or norm report is generated for a section of track or road, on which signage has been confirmed to be set-up correctly to current standards by traditional surveying methods, and subsequent test reports are compared against this datum or norm, to uncover sign misalignments, 1 5 1 7, A computer-implemented method for performing an alignment test on signage, comprising: receiving image or video data of signage from a camera mounted on a locomotive or vehicle; performing shape recognition on the image or video data so as to recognise shapes as target signs; calculating attributes of said target signs; and using the calculated attributes to generate output data indicative of the set-up oftarget signs within a given track or road section.18. A computer program comprising computer-readable instructions that, when executed on a processing system, perform the method of any of claims 10 to 17.