Automatic positioning of 2D image sign sightings in 3D space

1. A method for sign detection in a three-dimensional point cloud and a two-dimensional image, the method further comprising:

receiving point cloud data associated with a geographic region;

classifying at least one point neighborhood in the point cloud data as linear, planar, or volumetric;

selecting a sign position candidate having at least one point neighborhood classified as planar;

receiving image data associated with the geographic region;

calculating a sighting frustum from the image data;

performing a comparison of the sighting frustum to the sign position candidate having at least one point neighborhood classified as planar, wherein the comparison includes:

calculating an intersection of the at least one point neighborhood and the sighting frustum; and

calculating a coverage value for the intersection of the at least one point neighborhood and the sighting frustum;

identifying a plurality of locations for sign detection in response to the comparison; and

providing a location for the sign detection from the plurality of locations in response to the comparison.

2. The method of claim 1 , wherein the sighting frustum is calculated based on a location of an image sensor that collected the image data associated with the geographic region.

3. The method of claim 2 , further comprising:

receiving location data for a vehicle associated with the image sensor, wherein the sighting frustum is calculated based on the location data for the vehicle.

4. The method of claim 1 , wherein the coverage value (CV) is calculated based on an area of an intersection A (SUC) of the at least one point neighborhood and the sighting frustum and the area of the at least one point neighborhood A(C) according to

CV

=

A

⁡

(

S

⋂

C

)

A

⁡

(

C

)

.

5. The method of claim 1 , further comprising:

calculating a parsimony value for the intersection of the at least one point neighborhood and the sighting frustum.

6. The method of claim 5 , wherein the parsimony value (PV) is calculated based on an area of an intersection A(S∩C) of the at least one point neighborhood and the sighting frustum and the area of the sighting frustum A(S) according to

PV

=

A

⁡

(

S

⋂

C

)

A

⁡

(

S

)

.

7. The method of claim 1 , further comprising:

calculating a coverage value for the intersection of the at least one point neighborhood and the sighting frustum;

calculating a parsimony value for the intersection of the at least one point neighborhood and the sighting frustum; and

determining a focal overlap value as a minimum value between the coverage value and the parsimony value.

8. The method of claim 7 , wherein multiple focal overlap values are determined between a plurality of point neighborhood and the sighting frustum, the method further comprising:

ranking the multiple focal overlap values according an ascending order; and

selecting a highest focal overlap value from the ranked multiple focal overlap values.

9. The method of claim 8 , further comprising:

removing at least one other sign detection associated with the sighting frustum.

10. The method of claim 8 , further comprising:

providing the sign detection to a vehicle controller.

11. The method of claim 8 , further comprising:

providing the sign detection to a geographic database.

12. A non-transitory computer readable medium including instructions that when executed by a processor are configured to perform:

receiving point cloud data associated with a geographic region;

receiving image data associated with the geographic region;

performing a comparison of the point cloud data to the image data to identify a plurality of association sets;

determining a focal overlap value for each of the plurality of association sets;

ranking the plurality of association sets based on focal overlap value; and

selecting a location for sign detection from the ranking of the plurality of association sets.

13. The non-transitory computer readable medium of claim 12 , including instructions that when executed by the processor are configured to perform:

calculating an object lifetime value for the plurality of association sets; and

removing one or more association sets in response to the object lifetime value.

14. An apparatus for sign detection in a three-dimensional point cloud and a two-dimensional image, the apparatus comprising:

a memory configured to store instructions; and

a processor configured to execute the instructions to perform:

receiving point cloud data associated with a geographic region;

receiving image data associated with the geographic region;

performing a comparison of the point cloud data to the image data to identify a plurality of association sets; and

determining a focal overlap value for each of the plurality of association sets;

ranking the plurality of association sets based on the focal overlap value; and

selecting a location for sign detection from the ranking of the plurality of association sets.

15. The apparatus of claim 14 , wherein the point cloud data is LIDAR (light detection and ranging) data.

16. The apparatus of claim 14 , wherein the plurality of association sets are ranked according to numerical values.

17. The apparatus of claim 14 , further comprising:

selecting a location for sign detection based on the focal overlap value.