In this paper initially we propose a novel approach to classify handwritten characters based on a directional decomposition of the corresponding chain-code representation. This is alternative to previous transformations of the chain-codes proposed by the authors, namely the ordered and random decomposition of the bit-planes resulting from the binary representation of the chain-codes. Subsequently we utilize the power of the recently developed multiple classifier schemes using sntuple classifiers to integrate the complimentary information encapsulated in all three transformations into a more powerful and robust character recognition system. The results obtained through a series of cross-validation experiments show that the proposed fusion scheme not only outperforms its constituent parts and a number of other successful classifiers, but also enables significant savings in memory requirements compared to the original sntuple-based recognition system.