You cannot get 'the' equation, but it is possible to get an approximate or best fit, provided that you have the data points.
Look up 'curve fitting' in the manual or help section of Mathcad. Focus on functions like regress() or any of the xxxfit() functions.
You won't get THE equation. All you have is a number of points, so there are an infinite number of different equations possible. One of them could be a polynomial of n-1 th order if you have a datavector of n points. But of course you can also find logarithmic, exponential and other types of function or combinations of them which include all of your points. You have the choice between exact interpolation or approximation.
As valery wrote - attach you sheet, show what you have done so far and tell us as precise as possible what the intent is.
As far as we can see from the screenshot provided, Valentin does not need to digitize a plot as he already has the coordinates of the points in the matrix A. What he wants to find is an equation.
Werner Exinger wrote:
But of course you can also find logarithmic, exponential and other types of function or combinations of them which include all of your points.
See the attach here http://communities.ptc.com/message/152121#152121
From the screenshot Valentin provided it seems that he already has his points, he is searching THE(???) equation, not digitization.
Getting the data is the first order of business. The cited thread mentions, I think, a program called Tracer, which, on a good day, will trace a picture of a graph and provide an ordered list of coordinates for points on the curve. One version can be downloaded through here https://sites.google.com/site/kalypsosimulation/Home/data-analysis-software-1
I haven't tried this particular version; I generally use 1.7, because there were some annoyances in 2.0.0, which might have been fixed in this version.