Visual pathology is one of the leading health complications in present day. The number of individuals with retinopathy signs has increased significantly in the recent years. Timely medications have produced favorable results in evading the visual deficiency. However, the exploration of retinal photographs is another option for periodical retinal treatment which is non-invasive. Various approaches have presented for inspection of the retinal photographs which are based on supervised techniques that utilize important distinctive features to be extracted from good quality retina images for training. In contrast, unsupervised methods perform exploration of the retinal image without the utilization of any trained information or human intervention. These type of models have been developed in this research. The proposed framework identifies an eye objects (e.g. optic disc, fovea and macula) in the first step. Later, these objects along with the background are eliminated from the retinal photographs in order to attain the retinal vascular map to locate the lesion manifestation. Inspired from the current research of retinal blood vessels segmentation, this dissertation introduces new approaches for localization of blood vessels in retina images. This research is focused on solving problems of the tiny vessels detection, manual threshold selection, computation complexity, degradation of accuracy with the change of the dataset, improper handling of pathological images, contrast enhancement, erroneous classification of blood vessels and denoising of retina images. The different preprocessing steps, contrast enhancement techniques, thresholding schemes, vasculature enhancement schemes and postprocessing techniques of vessels segmentation in retina images have been explored in this dissertation. Five new methods have been introduced for localization of the vascular map in retina images. The first two methods modify the basic thresholding techniques along with strong preprocessing and postprocessing steps. A third method also includes thresholding for blood vessels segmentation along with a novel scheme for contrast enhancement that makes the vessels more prominent from the background. These methods localized both wide and tiny vessels accurately and automatically of the different datasets. These methods are also robust against noise and various retinal disease abnormalities. Additionally, reduction in the computation complexity of all the methods is observed. Experimental results of the proposed methods have accomplished remarkable improvement in terms of accuracy, sensitivity, specificity and area under the curve (AUC). The last two proposed methods introduced parallel steps for noise elimination, vasculature enhancement and segmentation. Initially, vessels location map (VLM) is extracted by utilizing various preprocessing and thresholding techniques. Then vasculature based enhancement is achieved by applying B-COSFIRE filter and Frangi’s filter in each method individually. In the first step of both methods, background noise and geometrical objects (e.g. optic disk, macula, fovea, etc.) are removed. Thus, VLM is obtained by identifying wide vessels positions to couple with the output image of the second step. The proposed methods have shown better accuracy, sensitivity and specificity when compared to existing methods. The proposed methods of this dissertation can indeed contribute to the development of an automatic retinal vessel segmentation system capable of handling both healthy and unhealthy images of all datasets.