Satellite image classification across multiple resolutions and time using ordering constraint among instances

1. A method comprising:

receiving a satellite image of an area;

classifying each pixel in the satellite image using a model;

for each of a plurality of possible water levels, determining a cost associated with the water level, wherein determining the cost associated with a water level comprises:

determining a number of pixels for which the model classification must change to be consistent with the water level; and

determining a difference between the water level and a water level determined for the area at a previous time point;

selecting the lowest cost water level; and

using the selected lowest cost water level to reclassify at least one pixel.

2. The method of claim 1 wherein determining the cost associated with a water level further comprises applying a weight to the determined difference between the water level and the water level determined for the area at a previous time.

3. The method of claim 2 further comprising selecting the weight to provide a lowest cost.

4. The method of claim 1 further comprising:

receiving a plurality of satellite images of the area;

beginning with a last satellite image in the plurality of images, recursively determining a water level for each satellite image of the area.

5. The method of claim 4 wherein recursively determining a water level for a satellite image comprises:

for each of a plurality of possible water levels, determining a cost associated with the water level, wherein determining the cost associated with a water level comprises:

determining a number of pixels for which a model classification must change to be consistent with the water level;

determining a difference between the water level and a water level for the area at a previous time point; and

determining a cost for the water level for the area at the previous time point; and

selecting the lowest cost water level.

6. A method comprising:

receiving classifications of a plurality of pixels that form an image, each classification indicating whether the pixel represents water or land;

for each of a plurality of possible water levels, determining a cost associated with the water level, wherein determining the cost associated with a water level comprises:

determining a number of pixels for which the classifications must change to be consistent with the water level; and

determining a difference between the water level and a water level determined for the area at a previous time point;

selecting the lowest cost water level; and

using the selected lowest cost water level to reclassify at least one pixel.

7. The method of claim 6 wherein determining the cost associated with a water level further comprises applying a weight to the determined difference between the water level and the water level determined for the area at a previous time.

8. The method of claim 7 further comprising selecting the weight to provide a lowest cost.

9. The method of claim 6 further comprising:

receiving a plurality of images of the area;

beginning with a last image in the plurality of images, recursively determining a water level for each image of the area.

10. The method of claim 9 wherein recursively determining a water level for an image comprises:

for each of a plurality of possible water levels, determining a cost associated with the water level, wherein determining the cost associated with a water level comprises:

determining a number of pixels for which a classification must change to be consistent with the water level;

determining a difference between the water level and a water level for the area at a previous time point; and

determining a cost for the water level for the area at the previous time point; and

selecting the lowest cost water level.