A 3D shape retrieval framework for 3D smart cities

This paper introduces the importance of 3D shape retrieval frameworks in 3D smart cities, and proposes a unified framework for 3D shape retrieval. The proposed method is based on the concept of visual representation, where each object is rendered with several depth images and binary images from distributed vertices in the regular polyhedron. For each image, several shape descriptors are utilized to extract features. Finally, different feature vectors are concatenated into a composite one. The experimental results show that the proposed method not only significantly improves the retrieval performance, but also achieves better retrieval effectiveness than other state-of-the-art algorithms, running princeton shape benchmark (PSB) and other standard evaluation measures.