Section Modulus calculator with image processing

I have a code that does the following:

1) I first plot a geometric shape with known dimensions (in this case it's an "I" shaped section of structural steel).

2) The plotted shape is captured from the figure and converted to binary.

3) Using the image processing tools of area and centroids, I calculate the plastic section modulus of the shape with the areas and centroid distances being in the units of pixels.

4) Because I know the width of the top part of the section, I am correlating it to its width in pixels in the image and using that to convert the section modulus from (pixels^3) to (in^3).

The problem I'm running into is that the hand calculated value does not match the converted value spit out by my code. The code value is on the order of several hundred units off from the actual value.

Stuff I've tried:

• I've used the perimeter of the section to try and get a conversion factor in pixels/inch, which didn't work.
• I've used the actual area of the section and correlated it to the area of the section in pixels, which didn't work.

Is there a way to somehow normalize the axes of the plotted shape (with actual dimensions) and then allow me to directly calculate the section modulus without needing to convert?

I should also note that I intend to iterate this to calculate section moduli of several deformed non-symmetric "I" shapes.

Thanks in advance.

[Daniel kiracofe]

I'd suggest that you post your code, with some sample inputs and the expected hand-calculated output. Without that I don't think I can be of much help to you.