Efficient and scalable DNA-based data storage requires encoding strategies that balance sequence compactness, stability, and fidelity. In this study, we present a randomized DNA base sequence design framework, designated RN-B#, which incorporates degenerate bases to significantly enhance information density and minimize sequence redundancy. By implementing rule-based encoding systems with varying constraints on homopolymer length and degenerate base positioning (e.g., R∞-B32, R2-B52, and R0-B16), we demonstrate the tunability of encoding properties such as GC balance, homopolymer suppression, and sequencing fidelity. Experimental validation using black-white binary image data encoded with RN-B# rules confirmed successful image recovery via Sanger sequencing, with an average sequence identity of up to 75%. Furthermore, we developed probabilistic models to quantify the sequencing accuracy as a function of sequencing depth and degenerate base complexity and corroborated them by in silico analysis. Our approach achieved a maximum theoretical information density of 3.91 bits/nt, offering a versatile platform for robust, high-capacity DNA data storage by leveraging the combinatorial space of degenerate nucleotide codes.