Dear B R Siddharth try substrate spectra without nanoparticles, and then the result subtracted from the spectra of nanoparticles. Not sure that this is methodologically correct, but it can work.
You really should explain better what your noise is. An example spectrum would help.
Low signal to noise in a UV/vis spectrum can be because your sample absorbance is too low (low signal), or alternatively too high, resulting in absorbance of all the light (high noise).
You can measure the substrate spectra which contains no nanoparticles and only your solution you are doing your research in it ( i.e buffer solution....) then you can repeat the test with your batch contains nanoparticles and also solution, then you have to subtract the final results with the primary one then it will give you the only nanoparticles spectra. then in order to remove the background effect, you need to find the lowest data in your final calculation, then subtract all your data from this number.
I hope my explanation is good enough to solve your problem.