08-12-2015 08:23 PM
Hello,
I am trying to fit an image 2d array by Nonlinear LM fitting. Checked everything. But it doesn't fit and shows error #23033. I guess also the extract fitting point VI doesnt extract datas.
Sorry, I am new to Labview so there may be many mistakes.
Please someone help me figure out the extract point vi?
Thanks!
B
08-12-2015 08:25 PM
one more attachment
08-13-2015 12:13 AM
The Non_linear_Lev-Mar_Fit_in_N_dimensions.llb is missing.
For the 2D non-linear fitting problem, you can refer to the following document.
08-13-2015 12:19 AM
There are many missing subVIs. You should also include some typical data.
None of your tools are from NI, so you should probably contact the author. He left his name on the front panel.
08-13-2015 07:51 AM - edited 08-13-2015 07:51 AM
@trrr : Sorry. Now attached the The Non_linear_Lev-Mar_Fit_in_N_dimensions.llb. How about now?
Could u have a look at it please?
@altenbach: Yes i will try to contact the person who wrote the code.
08-13-2015 09:42 AM
We still don't have any data. We cannot test unless we have typical input values.
Please create indicators for all data input wires to the fitting VI and run your VI until you get the error (and the indicators contains data)
Now stop the VI, right-click the new indicator terminals and select "create constant".
Now attach the VI once more.
We can now delete the rest of the code and run the fitting part under clean room conditions with the constants as inputs.
Are you sure that the problem is with the fitting and not with the earlier data prep? Do the dimensions match? Is the number of paramters correct, etc.
08-13-2015 10:16 AM
Error -23033 means:
Analysis: Vectors have different dimensions or empty vectors.
If you look at the fitting VI, you can see where this error is generated. Looks like it mostly depends on the output of "List of Variables", so start there..
08-13-2015 11:06 AM
In any case, you should really adapt my 2D code, it will be orders if magnitude faster using a subVI model instead of formula parsing. Look at the 2D gaussian fit in the link above. Just change the model slightly, the rest should fall in place.