The exponential growth of biological data, driven by high-throughput sequencing technologies, has created a pressing need for efficient and interpretable feature extraction methods in genomics. This study presents the BITSER method, a novel alignment-free approach for feature extraction and classification of biological sequences. Drawing inspiration from texture analysis techniques in computer vision, BITSER adapts the Local Binary Pattern (LBP) and its variants to extract histograms from sequences based on the electron-ion interaction potential (EIIP) of nucleotides. This method uses the raw FASTA sequences, avoiding the need for prior annotation or alignment, and offers a transparent and biologically meaningful feature representation. The BITSER was evaluated on SARS-CoV-2 and DENV viral genomes, achieving classification accuracies exceeding 99% across multiple classifiers by adopting different classification algorithms. Feature importance analysis enabled significant dimensionality reduction, preserving predictive power with only a few features per dataset. Compared to existing methods, BITSER demonstrated superior results, computational efficiency, and interpretability. These results underscore the method’s robustness, scalability, and potential for broad application in large-scale genomic analysis.