Hyperspectral imaging offers a significant edge over standard RGB and multispectral images for land classification. It captures a wider range of electromagnetic waves, producing more detailed images than previous methods. This allows objects to be identified and distinguished with high certainty due to hyperspectral capabilities. However, the large data volume makes reducing the computational workload challenging. Imbalanced data and suboptimal hyperparameter settings can reduce classification accuracy. Hyperspectral image classification is computationally demanding, especially with mixed-pixel issues in high-resolution images. This study uses EO-1 satellite imagery with a 30-meter resolution affected by mixed pixels. It introduces a new classification approach to effectively use hyperspectral remote sensing at this resolution. The process includes satellite image preprocessing—geometric correction, image enhancement with FLAASH, and geometric and atmospheric corrections. To lessen the computational burden, a slab approach partitions the 242 spectral bands into segments, extracting features from each, resulting in fewer total features. These features are then input into a support vector machine (SVM) for five-class classification. Parameters like polynomial order, kernel scale, and kernel type are tuned for optimal accuracy. A novel SLAB Multi-Fold technique is proposed. Results indicate that the slab method combined with SVM achieves a maximum accuracy of 51.39%. The best results came from slab 2, with a polynomial order of 8 and k=4, using both linear and Gaussian kernels. These findings offer valuable insights for future research on satellite image classification, especially when tuning multiple hyperparameters within this SLAB approach. Future work could compare these results with higher-resolution images and different datasets to better evaluate the technique's accuracy.