International Journal of Computer Applications |
Foundation of Computer Science (FCS), NY, USA |
Volume 27 - Number 3 |
Year of Publication: 2011 |
Authors: S.Anbumalar, P.Rameshbabu, R.Anandanatarajan |
10.5120/3281-4469 |
S.Anbumalar, P.Rameshbabu, R.Anandanatarajan . Chromatograms separation using Matrix decomposition. International Journal of Computer Applications. 27, 3 ( August 2011), 24-32. DOI=10.5120/3281-4469
Non – negative matrix factorization (NMF) was generally used to obtain representation of data using non – negativity constraints. It lead to parts – based (or) region based representation in the vector space because they allow only additive combinations of original data. NMF has been applied so far in image and text data analysis, audio signal separation, signal separation in bio-medical applications and spectral resolution. The original Lee and Seung ‘s NMF has to be modified for chemical analysis, based on the characteristics of that signal. In this paper, sparse NMF (sNMF) has been used for the deconvolution of overlapping chromatograms of chemical mixture. Before applying sNMF, the number of components in mixture was determined using Principal Component Analysis (PCA). The experimental overlapping chromatograms were obtained using Gas Chromatography –Flame Ionization Detector (GC-FID) for the chemical mixture of acetone and acrolein and they have been soundly resolved by sNMF algorithm. The proposed algorithm has also been tested with simulated two, three and four component chromatograms of severely overlapped and embedded peaks. Even though there are three or four components, the results are encouraging. The correlation coefficient is greater than 0.99 and signal to noise ratio is greater than 29dB always.