IMAGINE Discussions

[Robert1968]

So when working in AutoSync, sometimes I like to look at the Preview Output to check for control points from A to B..

I am trying to understand how this works.. it gets confuse at sometimes... I would like to get some extra information

Can anyone out there help me to get me understand how it works ? 

Thanks,

Robert 

[ian.anderson · Accepted solution]

Hi Robert,

What you're effectively seeing is the error vector - the difference between where the control point should be in map space and the location the solved model predicts it to be to fit the model. I.e. it's a measure of how well the computed/solved geometric model fits the original points. Big error vectors indicate a poor fit. 

For example, AutoSync might generate a point at 100000, 100000 (along with a whole bunch of other points). It then solves the geometric model you have asked for using all these points. It then feeds the points back through the model to see where the model predicts those points should fall. The model might predict that point to be at 100010, 100010. So at that point there's a (potential) error of 14.1m. You might decide that's acceptable or you might want to check out the points with the largest errors and modify them or delete them entirely. Then resolve the model using the updated points. 

To help evaluate the points you can also switch the top image on/off (or use Swipe, etc) to see how well the predicted image model fits over the reference imagery. If you see big distortions then you may need to edit the points (or pick a different model).

Hope that helps.

Cheers

[Robert1968]

Thank you for respond to my question here...

What exactly do you mean by 100000, 100000 ? Do you mean 100000, 100000 points around the image before you run the APM ?

[sbent]

Ian is using 100000, 100000 as an example of a X,Y coordinate where AutoSync generated a point. 100010 (X), 100010 (Y) is the predicted point coordinate.

Regards,

Stephen Bent

Hexagon Geospatial Support