You have two concerns, which will both be influenced by your location.
First is a possible copy-right violation. Since what you are doing constitutes fair-use and even if it would not, the copy-right holder to the image would probably not enforce copy-right (no gain to be shared, damage to copy-right holder hard to ascertain, use is innocent, enforcing copy-right would be against academic customs, ...), you are safe. You do not need the author's permission, and probably, by now, the copy-right would rest with a third party such as the editor of a book or the journal in which the text and image were published.
The second concern is an accusation of plagiarism. Standards have shifted considerably over the last decades. A vague citation of the image would have once been sufficient, but you need to be a bit more careful these days. If your image is derived from the TEXT, then it is more than sufficient to put in the caption of the image something on the line
Image after [4, Ch.3]
If you feel that the image benefits a lot from the actual figure, it would be sufficient to cite the figure
Image after [4, Fig.23(a)]
where of course you adjust the citation style. If the function of the figure is crucial to your point, you might even want to say something like:
"We explain and illustrate the concept of a shock absorber along the lines of Nikolas Bourbaki's Cohomology of Springs [4]. Bourbaki's argument first considers an algebraic sheaf made out of corrugated steel and its interaction with a source of friction. We illustrate this concept with Figure 2, which is a strongly adapted version of Fig. 23 in [4]."
The last one would certainly be overkill unless you are just rephrasing Bourbaki (a famous, but ill-fated general from the Prusso-French war in 1871) who to my knowledge never worked on spring theory) and the argument is not standard.
The point is that you should not incorporate other's thoughts and results without proper attribution. For almost all purposes, a single citation in the caption of the image is sufficient, where you cite either the master image or the text, depending on the source of your illustration. You do not need the author's permission for doing that.
(If however the original image would have artistic value, then things are different. For instance, do not redraw an XKDC cartoon.)
In short: you are almost certainly over-thinking, you do not need permission to use someone's scientific / engineering illustration, and you need to attribute material that you used directly or indirectly.