This has been discussed before, but it was 4 years ago so I thought it would be interesting to get an update.
The threshold method is limited to 8 or 32 bit images. However, looking at the code today it is apparent that most of the datatypes that are supported in OpenCV have been implemented in thresh.cppbut that the public API does not reflect the available functionality.
This is more than a little frustrating. Especially if you are relying on a wrapper such as Emgu which only provides the functionality that the opencv API documentation exposes.
Does anyone have any insight in why the added depth support has not been exposed and if there is any chance for things to improve in 4.0? Or should I just accept fact and work around this arbitrary limitation as best I can.