METHOD, APPARATUS, AND SYSTEM FOR A PARAMETRIC REPRESENTATION OF SIGNS

• US 20180300564A1
• Filed: 04/13/2017
• Published: 10/18/2018
• Est. Priority Date: 04/13/2017
• Status: Active Grant

• Current Assignee: HERE Global B.V. (Nokia Corporation)
• Original Assignee: HERE Global B.V. (Nokia Corporation)
• Inventors: KWANT, Richard, MITTAL, Anish, LAWLOR, David, GUPTA, Himaanshu
• Granted Patent: US 10,325,166 B2
• Time in Patent Office: Days
• CPC Class Codes: B60W 2420/403 Image sensing, e.g. optical camera
• CPC Class Codes: B60W 2554/00 Input parameters relating to objects
• CPC Class Codes: B60W 2555/60 Traffic rules, e.g. speed limits or right of way
• CPC Class Codes: B60W 30/0956 the prediction being responsive to traffic or environmental parameters
• CPC Class Codes: B60W 40/04 Traffic conditions
• CPC Class Codes: G06F 18/23 Clustering techniques
• CPC Class Codes: G06V 10/44 Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
• CPC Class Codes: G06V 10/46 Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features colour feature extraction G06V10/56
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
• CPC Class Codes: G06V 10/56 relating to colour
• CPC Class Codes: G06V 20/582 of traffic signs

Abstract

An approach is provided for parametric representation of signs. The approach involves receiving a request to detect and encode signs depicted in an input image into a parametric representation. The approach also involves assigning processing nodes of a computer vision system to independently process each grid cell of the input image to detect at least one edge of a sign. The processing nodes are assigned based on proximity to each grid cell. Each respective grid cell is created by overlaying a grid onto the input image. The approach further involves encoding, by the processing nodes, an angle and a location of a detected edge as edge parameters of a cell-based parametric representation for each grid cell. The approach further involves aggregating the cell-based parametric representation for each respective grid cell in which at least one edge is detected to output the parametric representation of the at least one sign.

Claims

1. A method comprising:

receiving, by a computer vision system, a request to detect and encode at least one sign depicted in an input image into a parametric representation of the at least one sign;

assigning respective one or more processing nodes of the computer vision system to independently process each respective grid cell of the input image to detect at least one edge of the at least one sign, wherein said processing nodes are assigned based on a proximity to said each respective grid cell, and wherein said each respective grid cell is among a plurality of grid cells created by overlaying a grid onto the input image;

encoding, by said respective processing nodes, an angle and a location of a detected at least one edge as edge parameters of a cell-based parametric representation for said each respective grid cell; and

aggregating the cell-based parametric representation for said each respective grid cell in which the at least one edge of the at least one sign is detected to output the parametric representation of the at least one sign.

2. The method of claim 1, wherein there are more than one of said processing nodes assigned to process the respective grid cell to generate a plurality of redundant cell-based parametric representations for said each respective grid cell, and wherein the plurality of redundant cell-based parametric representations are processed into a single cell-based parametric representation to represent the cell-based parametric representation for said each respective grid cell.

3. The method of claim 1, further comprising:

determining a confidence value associated with the detected at least one edge,

wherein the confidence value is further encoded into the cell-based parametric representation.

4. The method of claim 1, further comprising:

determining a reference point and a reference angle for said each respective grid cell;

encoding the location of the detected at least one edge as a distance along a line segment drawn from the reference point to the detected at least one edge, wherein the line segment is perpendicular to the detected at least one edge; and

encoding the angle of the detected at least one edge in relation to the reference angle.

5. The method of claim 4, wherein the reference point is a centroid of an area of said each respective grid cell that represents image data corresponding to the at least one sign.

6. The method of claim 4, wherein the cell-based parametric representation includes a plurality of output channels, and wherein each of the plurality of output channels corresponds to an edge detector that is responsible for a different range of edge angles.

7. The method of claim 1, further comprising:

determining a predicted sign center for said each respective grid cell in which the at least one edge of the at least one sign is detected,

wherein the cell-based parametric representation further includes a parameter indicating the predicted sign center.

8. The method of claim 7, further comprising:

clustering the cell-based parametric representation for said each respective grid cell in which the at least one edge of the at least one sign is detected into one or more clusters based on the predicted sign center for said each respective grid cell; and

designating each of the one or more clusters as representative of a different sign when the input image depicts more than one sign.

9. The method of claim 7, wherein a parameter representing the predicted sign center is indicated as an X, Y displacement from said each respective grid cell to the predicted sign center.

10. The method of claim 1, further comprising:

processing the cell-based parametric representation for said each respective cell in the parametric representation to reconstruct one or more line segments corresponding to the detected at least one edge cell; and

joining the line segments to reassemble a polygonal representation of the at least one sign.

11. The method of claim 1, wherein cell-based parametric representation further includes an attribute parameter, and wherein the attribute parameter indicates one or more characteristics of the detected at least one edge, the at least one sign, or a combination thereof.

12. An apparatus comprising:

at least one processor; and

at least one memory including computer program code for one or more programs,

the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus to perform at least the following,

receive, by a computer vision system, a request to detect and encode at least one sign depicted in an input image into a parametric representation of the at least one sign;

assign respective one or more processing nodes of the computer vision system to independently process each respective grid cell of the input image to detect at least one edge of the at least one sign, wherein said processing nodes are assigned based on a proximity to said each respective grid cell, and wherein said each respective grid cell is among a plurality of grid cells created by overlaying a grid onto the input image;

encode, by said respective processing nodes, an angle and a location of a detected at least one edge as edge parameters of a cell-based parametric representation for said each respective grid cell; and

aggregate the cell-based parametric representation for said each respective grid cell in which the at least one edge of the at least one sign is detected to output the parametric representation of the at least one sign.

13. The apparatus of claim 12, wherein there are more than one of said processing nodes assigned to process the respective grid cell to generate a plurality of redundant cell-based parametric representations for said each respective grid cell, and wherein the plurality of redundant cell-based parametric representations are processed into a single cell-based parametric representation to represent the cell-based parametric representation for said each respective grid cell.

14. The apparatus of claim 12, wherein the apparatus is further caused to:

determine a confidence value associated with the detected at least one edge,

wherein the confidence value is further encoded into the cell-based parametric representation.

15. The apparatus of claim 12, wherein the apparatus is further caused to:

determine a reference point and a reference angle for said each respective grid cell;

encode the location of the detected at least one edge as a distance along a line segment drawn from the reference point to the detected at least one edge, wherein the line segment is perpendicular to the detected at least one edge; and

encode the angle of the detected at least one edge in relation to the reference angle.

16. The apparatus of claim 12, wherein the apparatus is further caused to:

determine a predicted sign center for said each respective grid cell in which the at least one edge of the at least one sign is detected,

wherein the cell-based parametric representation further includes a parameter indicating the predicted sign center.

17. A non-transitory computer-readable storage medium, carrying one or more sequences of one or more instructions which, when executed by one or more processors, cause an apparatus to perform:

receiving, by a computer vision system, a request to detect and encode at least one sign depicted in an input image into a parametric representation of the at least one sign;

assigning respective one or more processing nodes of the computer vision system to independently process each respective grid cell of the input image to detect at least one edge of the at least one sign, wherein said processing nodes are assigned based on a proximity to said each respective grid cell, and wherein said each respective grid cell is among a plurality of grid cells created by overlaying a grid onto the input image;

encoding, by said respective processing nodes, an angle and a location of a detected at least one edge as edge parameters of a cell-based parametric representation for said each respective grid cell; and

aggregating the cell-based parametric representation for said each respective grid cell in which the at least one edge of the at least one sign is detected to output the parametric representation of the at least one sign.

18. The non-transitory computer-readable storage medium of claim 17, wherein there are more than one of said processing nodes assigned to process the respective grid cell to generate a plurality of redundant cell-based parametric representations for said each respective grid cell, and wherein the plurality of redundant cell-based parametric representations are processed into a single cell-based parametric representation to represent the cell-based parametric representation for said each respective grid cell.

19. The non-transitory computer-readable storage medium of claim 17, wherein the apparatus is further caused to perform:

determining a confidence value associated with the detected at least one edge,

wherein the confidence value is further encoded into the cell-based parametric representation.

20. The non-transitory computer-readable storage medium of claim 17, wherein the apparatus is further caused to perform:

determining a reference point and a reference angle for said each respective grid cell;

encoding the location of the detected at least one edge as a distance along a line segment drawn from the reference point to the detected at least one edge, wherein the line segment is perpendicular to the detected at least one edge; and

encoding the angle of the detected at least one edge in relation to the reference angle.