Minutiae-based matching is commonly used in fingerprint recognition systems due to its proven performance. However, such matching procedure usually involves unordered and variable size templates and it does not favour emerging bio-cryptography applications and most classifiers. This paper proposes a solution by converting the original minutiae set into a bit-string through the amalgamation of bag-of-words modelling, multi-scale construction and dynamic quantization. Experimental results show that the proposed method has high potential in biocryptography applications due to its outstanding EER of 0:51\% and entropy of 723 bits. Further security and privacy concerns are also analyzed.