Glaucoma is the stern ailment that affects the eye and causes blindness without screening significant symptoms at early stage. Glaucoma is instigated due to unseemly draining of aqueous humour within the eye. The eye diseases such as glaucoma, diabetic retinopathy are diagnosed using the retinal features such as Optic Disc (OD) and Optic Cup (OC). In this paper, the optic disc region is segregated from the input retinal image using Local Region Recursive Segmentation (LRRS). The discriminative features are extracted from segmented optic disc using the proposed LP_LDS method. In this LP_LDS method, the micro level details of optic disc are extracted by considering the pairwise directions of vector of the referenced pixel and its neighbourhood using Local Vector Pattern (LVP). The resultant image after extraction is decomposed with 2D Little Wood Paley (LP) Empirical Wavelet Transform to obtain the detailed sub images. The sub images are disintegrated and their features are normalized using z-score normalization. Finally obtained features are classified using different classifiers. Among the four classifiers (Support vector machine classifier, K-Nearest Neighbour classifier, Random Forest Classifier, Decision tree classifier), Random Forest Classifier exhibits better performance than other classifiers. This LP_LDS method has been examined using MIAG RIMONE (Release2) database that contains 255 normal images and 200 glaucoma images. This LP_LDS method achieves a sensitivity of 89.29% in classifying glaucoma images and an overall accuracy of 89.71% using Random Forest Classifier.