Introduction: The human eye is the most often affected by Diabetes. It deteriorates the functioning of the retina. Diabetic Retin opathy, Glaucoma, Macular Edema are some of the commonly found ophthalmic disorders among diabetic people. Microaneu rysms are the earliest symptoms of Diabetic Retinopathy. If these symptoms are detected and treated in the early stages, then they can avoid vision loss among patients. Objective: The main objective of this research work is to detect microaneurysms present on the surface of the retina that char acterizes Non-Proliferative Diabetic Retinopathy. Methods: The algorithm proposed in this paper is based on morphological operations eliminating blood vessels and the optic disc, followed by the detection of Microaneurysms. Grey level Co-occurrence Matrix(GLCM) and Histogram of Oriented (HOG) features are extracted to identify the Microaneurysms, and various classifiers are tested. HOG features are useful in finding the regions affected in a retinal image as they compute the histogram of gradients that can facilitate tasks such as classification, detection, and recognition. Various classifiers are tested on public fundus databases with the proposed method. Results: For the STARE Database, the Decision tree classifier yielded values for accuracy 0.95, recall 0.9, and precision 1.0. Similarly, accuracy 0.91, recall 0.94, and precision 0.92 were computed for the thee-Ophtha database and accuracy 1.0, recall 0.94, and precision 1.0 for the DIARETDB1 database. Conclusion: From experimentation, it was observed that the decision tree classifier and HOG features yield the best results for detection of Non-Proliferative Diabetic Retinopathy(NPDR) from most of the public fundus databases. The encouraging values computed for performance evaluation testify to the efficiency and robustness of the proposed method.