Diabetic retinopathy (DR) occurs due to damage caused by elevated blood sugar levels to the small blood vessels in the retina, leading to vision impairment or blindness if not detected early. With the global rise in diabetes cases, early detection of DR has become increasingly important for preventing severe complications. This paper introduces a novel approach for the automatic identification of DR using Zernike moment decomposition applied to retinal fundus images. The methodology involves preprocessing fundus images from some open datasets with Contrast Limited Adaptive Histogram Equalization (CLAHE) and Gaussian image filters to enhance image quality. Feature extraction is performed using shape descriptors, followed by a subsampling technique to address imbalanced multiclass distributions. A Support Vector Machine (SVM) classifier is then employed to automatically detect DR. Results indicate that the model achieves high sensitivity (0.90) and specificity (0.89) in detecting non-DR cases. However, improving sensitivity in identifying mild and proliferative DR stages remains a challenge for further development.