Coronary artery disease is one of the leading cause of death all over the world in past years especially in industrial countries. Among all of the different ways for diagnosis this disease X-Ray angiograms known as gold standard for cardiologist to diagnosis this heart disease and decide if angioplasty surgery is needed. X-Ray angiogram images usually contains noises and have poor quality and low contrast between vessel and non-vessel area. This reason made vessel enhancement and vessel extraction an important task in recent years. Many rule based and learning based approaches are proposed for vessel segmentation. In this thesis we proposed two rule based and one learning based vessel extraction algorithm. The first method a new vessel and centerline extraction method based on vessel tracking approaches. The second one is new coronary arteries segmentation method based on circular flux-flow measuring of angiogram images. And the last proposed method is a vessel extraction learning algorithm based on using hierarchy structure of neural network. Finally the proposed methods are evaluate by some existing evaluation criteria such as accuracy, sensitivity, and specificity and we compare them with two other approaches. کلمات کلیدی : 1. Coronary artery disease 2. X-Ray angiography 3. Vessel tracking 4. Circular flux flow measurement 5. Neural Network 6. Image processing