The spectrum profiles you attached have a problem with the wavelength calibration which the graphs you posted do not so I went back to your spectrum image and recalibrated that in wavelength first. I then calculated the instrument response and applied it to the raw spectrum to produce a final instrument response corrected spectrum. You can see below the raw spectrum (blue), the instrument response (orange) and the final result (green).
As a check I also compared the final result (green) with the reference spectrum (pink) which I filtered to match the resolution of the measured spectrum. You can see the good agreement in the shape of the continuum which confirms the instrument response is working correctly.
I have attached the steps I took to produce the final spectrum in the attached pdf. I used Vspec as it is easier to see the individual steps but you can use any software package to produce the same result.
I hope you can follow this ok