In the digital world, two-dimensional (2D) and three-dimensional (3D) shapes are important for representing real objects. Their applications span a wide range of fields, including medical, engineering, and security, etc... Considering the aspect that 2D and 3D models are widespread and because graphs are strong mathematical modeling tools used in a variety of computer science domains. We aim to represent our input as graphs to benefit from the highly meaningful representation. In this thesis, we conduct two parts.The first part was related to the 3D models, where we addressed the problem of finding a superior one-to-one correspondence between the 3D models to obtain an optimal matching and retrieval. To do so, we detect feature points using...