Abstarct: One of the most important challenges in biometric recognition area is finding new recognition methods with more applicability and proper functionality. Because of its importance in human recognition capabilities, face recognition is attended as a convenient case for this purpose. With respect to 3D nature of face, 3D face recognition methods have more potential than 2D methods in solving existing problems. In addition, because of their capability in face synthesis and their lack of need to 3D input images, model baxsed 3D face recognition methods, seem to be more exciting and more important. In this research, after a literature review of face recognition methods, advantage and disadvantages of each of their branches would be studied. Because of wide application of computer graphics in 3D face recognition, this subject has been attended carefully and completely. Model baxsed 3D face recognition is divided into two main methods. The first one is an approach baxsed on generic models and the second method applies morphable models. For creating a morphable model, we need a 3D face databaxse. Because of lack of such a databaxse, we decided to make a 3D face models databaxse from 2D images. So, the photogrammetry algorithm which is baxsed on generic models, has been used. Hence, both main methods of model baxsed 3D face recognition have been studied and implemented in this thesis. In this research two 3D face databaxses have been created from 2D images. In first one, 2D face images are extracted from MPI databaxses, and in second one, face images of different Iranian races have been used. Because of its remarkable features and advantages, face recognition baxsed on morphable model is studied as the most important 3D face recognition method. Indeed, morphable model is a combination of deformable models and computer graphics techniques for simulation of illumination effects. In this model changes in face pose and illumination is implemented simply. Also this model can estimate pose, shape, texture and illumination of face model carefully and with no need of operator's attendances. In addition face synthesis and capability of facial exxpressions implementation on synthesized model, are of other advantages of this method. The core step of morphable modeling is finding a dense point to point correspondence between instant models. For implementing this step, in this research, a novel and graph baxsed method have been presented which optimizes image matching results, optimization results of this method, show remarkable improvement in morphable model functionality and more quality in face synthesis. This algorithm is a graph optimization algorithm. So it is independent of matching method and it can be implemented on results of all image matching methods. In addition to optimization of matching results, injectivation is one of the other benefits of this novel method. After implementation of morphable model, effect of 3D face databaxse on recognition and synthesis quality was studied. So, input images of Iranian faces, was synthesized by two morphable models baxsed on the two presented 3D databaxses. High quality of images synthesized by the morphable model baxsed on Iranian 3D face databaxse, shows effective role of databaxse in morphable modeling. In this research, faces were recognized by rate of 100% by the morphable model algorithm. In spite of low number of input test images, this high recognition rate, shows usefulness and power of this model in face recognition.