The 3D object recognition stands a very important role in the process of computer vision. The recognition of objects is fundamental step in the design of more automated systems. The need for these systems is now high. Therefore, the 3D object recognition has a high impact. The most frequently used methods are based on neural networks (NNs), or, the derivation of NNs. The NNs approaches have high impact on computer performance and the behavior of these systems is not easily predictable over time - this attribute can be seen in all chaotic systems. Sometimes it is very difficult to determine the optimal configuration of these systems and several tests are needed to figure out the optimal system configuration. The process is not very user friendly. We realize, the grammar approaches are more manageable and more predictable than NNs. Therefore, in this paper, we present the symbolization process. It is the first step of 3D object recognition based on grammar. Our aim is to describe the process allowing the 3D object symbolization.