Shape Description for 3D Model Retrieval

Advanced design and scanning technologies facilitates the process of creation of 3D models. Still, designing a 3D model from scratch remains an effort and time consuming task. At the same time, the growth of the amount of 3D digital data available on the Internet and in corporate repositories gives the possibility to retrieve and reuse already existing models. The natural way to search for textual information is verbal description, whereas this can be ambiguous for digital shapes. Recently research in the field of shape representation and analysis proposes several solutions for shape description. Being exploited in different application domains like Industrial Design, Architecture, Medical Imaging and Game Industry shape descriptors might possess different properties important for shape retrieval. Some of them are robustness to noise, invariance to model’s posture and position in space, ability to capture overall and partial similarity. At the same time it is important to have a compact shape descriptor to allow efficient retrieval process. Therefore there exists a tradeoff between completeness and compactness of a shape descriptor. The above-stated issues form a backbone of this PhD research which deals with description and retrieval of 3D models.