A new peak-based quantitative spectra matching method using affinity matrix spectral analysis is proposed. An affinity matrix is generated to describe similarity between 2D HSQC peaks given their spatial location. Application of the approach gives best matched peaks of two different spectra with the same or different number of peaks. Results were validated against a previously published method. The new algorithm was found to be an order of magnitude faster in finding matched peaks, in comparison to our previously reported technique. This speed-up allows large number of peaks to be matched quickly, meaning that structures of high complexity can be compared across large structure databases. Matching examples are provided for 15N HSQC spectra of protein samples, which we use to establish performance.