Long-tracker translation

[tdivoll]

Lots of code in the original MATLAB script; much may not be needed.

Need to look at start day and next day.

@danielmwatkins can you add a description of the minimum functionality needed?

[danielmwatkins]

Rather than translating MATLAB script, the plan is to work within the Julia code to develop methods for extending the image search to multiple floes. A key first step is the addition of floe labels.

I think the minimum functionality (which is probably doable on our end) is something like this:

max_days = (something) max_distance = (max_speed * elapsed_time)

for floe in (image such that floe is not matched):
    for next_image in (images with time < max_days):
       for floes within max_distance:
            look_for_matches(floe, next_image_floes)

So in order to do this, we'd need to

• make sure that the dates of images are available to the floe tracker module, which was enabled by the addition of floe labels
• see how to add parameters to the config file for the search limitations
• make sure that the documentation is clear enough on how the tracker module goes through the set of images so we know where to add the additional functionality

[cpaniaguam]

352

[danielmwatkins]

Coming back to this, we do really need a way to extend floe tracking past small gaps from cloud cover and/or from segmentation errors. I'm not sure whether this is best addressed in the Julia package or in the pipeline, though.