Error when using decsg (PDE Toolbox)

Hi there,

I recently encountered a problem with the function "decsg.m" of the PDE Toolbox. I normally use this function to generate a FE mesh after segmenting an image. This script has always worked.

I just rerunned some old data, and I got this error:

""Error in decsg (line 103) if isempty(gd), dl1=[]; bt1=[]; dl=[]; bt=[]; msb=[]; return, end

Output argument "dl1" (and maybe others) not assigned during call to "D:\USlib\decsg.m>decsg"."

This is quite weird and very surprising. I checked the input variables, they are all ok. No conflicts ... what's wrong here?

Kind regards

Richard

Answer 1

[Martijn]

Please make sure that your input matrix is valid. For example if it defined polygons, please make sure it does not contain any duplicate points.

[Richard]

Thank you Martijn, this is true: I added the starting coordinates of my polygons at the end to ensure periodicity ... everything works fine now! Sorry for the inconvenience and thank you all for your thoughts and for the help!

Answer 2

[Friedrich]

Hi,

is the variable gd (first input to the descsg function) empty? I am wondering about the location of the decsg.m function. Normally it should be in a subfolder called:

toolbox\pde\decsg.m

Can you run which -all decsg to see if you maybe overlaoded that function?

Answer 3

[Richard]

>> which -all decsg C:\Program Files\MATLAB\R2011b\toolbox\pde\decsg.m

gd is not empty, it's a 204 x 2 matrix

with row 1: 2's (= POLY) with row 2: N (= 101, number of points polygon) with row 3 - 103 : x-coordinates (N = 101) with row 104 - 204 : y-coordinates (N = 101)

colom 1 = P1 (total area) colom 2 = P2 (mid-area, which is substracted)

[Friedrich]

Okay thats strange. Line 103 shouldnt be any trouble maker. If you can reproduce that, I would contact MathWorks and attach the input variables as mat file to your service request. They should now whats going on here and provide you a solution.

[Bill Greene]

Hi Richard,

This is indeed very strange. Since it appears MATLAB is executing

the correct decsg function in your installation, I'm wondering if

the right version of isempty is getting executed.

If you type

which -all isempty

you should get a long list. Do any of these listed entries

look like they might be locally-defined versions?

One more test you might try is to set a break point at the

line in your function where decsg is called. Then, when the

MATLAB debugger stops there during execution, step into decsg.

The file that is opened should be the same as output from your

which -all decsg.

Then try to step into isempty. This should be a MATLAB builtin

so you shouldn't be able to step into this function.

Maybe these tests will give you some clues.

Regards,

Bill