Diabetic Retinopathy (DR) is a prevalent complication of diabetes, leading to significant vision loss and blindness if left untreated. Early detection is critical to effective DR management, enabling timely intervention to prevent visual impairment. This study focuses on implementing a feature extraction algorithm specifically designed to classify DR stages using retinal imaging data. The algorithm systematically extracts essential retinal features such as exudates, microaneurysms, hemorrhages, etc. For evaluation, the experimentation applied and tested the feature extraction algorithm across several well-known datasets to ensure its reliability and adaptability across diverse data sources. Comparative analyses demonstrated that the proposed method significantly enhances classification accuracy, offering clear improvements over existing approaches. By enhancing the precision of DR detection, this algorithm not only contributes to more reliable screening practices but also aids in the development of scalable, cost-effective solutions for managing diabetic retinopathy.