Cloud resilient algorithm for coastline registration

1. A method of image navigation of an input image by edge registration, the method comprising:

generating a reference image kernel having borders embracing edges that conform to edges of a reference image and image intensities of the reference image kernel internal to the borders increase with decreasing distance from the edges embraced thereby, the reference image indicating a geographic location of a feature therein;

computing a registration score by cross-correlating the image intensities of the reference image kernel with image intensities of edges in the input image;

assigning the geographic location of the feature in the reference image to the feature in the input image responsive to the registration score meeting a registration criterion; and

storing the input image of the feature with the geographic location thereof in a memory device.

2. The method of claim 1 , wherein generating the reference image kernel comprises:

applying a gradient filter to the edges of the reference image; and

subtracting the image intensities of the filtered edges from the image intensities of the edges to form the borders of the reference image kernel.

3. The method of claim 2 further comprising:

retrieving truth data corresponding to the feature in the input image; and

generating the edges of the reference image from the truth data.

4. The method of claim 3 , wherein retrieving the truth data comprises retrieving, as the truth data, an image of shoreline vector data.

5. The method of claim 3 , wherein retrieving the truth data comprises retrieving, as the truth data, a previously navigated image.

6. The method of claim 1 further comprising:

extracting the edges from the input image; and

normalizing the extracted edge data prior to cross-correlating the image intensities of the reference image kernel with the image intensities of edges in the input image.

7. An image processing apparatus comprising:

a database storing data from which a reference image is generated, the data including a geographic location of a feature in the reference image;

a processor configured to:

generate a reference image kernel having borders embracing edges that conform to edges of the reference image and image intensities internal to the borders of the reference image kernel increase with decreasing distance from the edges embraced thereby;

compute a registration score by cross-correlating the image intensities of the reference image kernel with image intensities of edges in an input image;

assign the geographic location of the feature in the reference image to the feature in the input image responsive to the registration score meeting a registration criterion; and

store the input image of the feature with the geographic location thereof in a memory device.

8. The apparatus of claim 7 , wherein the processor is further configured to:

apply a gradient filter to the edges of the reference image; and

subtract the image intensities of the filtered edges from the image intensities of the edges to form the borders of the reference image kernel.

9. The apparatus of claim 8 , wherein the processor is further configured to:

retrieve truth data corresponding to the feature in the input image; and

generate the edges of the reference image from the truth data.

10. The apparatus of claim 9 , wherein the processor is further configured to retrieve, as the truth data, shoreline vector data.

11. The apparatus of claim 9 , wherein the processor is further configured to retrieve, as the truth data, a previously navigated image.

12. The apparatus of claim 7 , wherein the processor is further configured to:

extract the edges from the input image; and

normalize the extracted edge data prior to cross-correlating the image intensities of the reference image kernel with the image intensities of edges in the input image.

13. A tangible, non-transient medium having instructions encoded thereon that, when executed by a processor, cause the processor to:

generate a reference image kernel having borders embracing edges that conform to edges of a reference image and image intensities internal to the borders of the reference image kernel increase with decreasing distance from the edges embraced thereby, the reference image indicating the geographic location of a feature therein;

compute a registration score by cross-correlating the image intensities of the reference image kernel with image intensities of edges in the input image;

assign the geographic location of the feature in the reference image to the feature in the input image responsive to the registration score meeting a registration criterion; and

store the input image of the feature with the geographic location thereof in a memory.

14. The tangible, non-transient medium of claim 13 having additional instructions encoded thereon that, when executed by the processor, cause the processor to:

apply a gradient filter to the edges of the reference image; and

subtract the image intensities of the filtered edges from the image intensities of the edges to form the borders of the reference image kernel.

15. The tangible, non-transient medium of claim 14 having additional instructions encoded thereon that, when executed by the processor, cause the processor to:

retrieve truth data corresponding to a landmark being sought in the input image; and

generate the edges of the reference image from the truth data.