Incomplete 3D Shape Retrieval via Sparse Dictionary Learning

How to deal with missing data is one of the recurring questions in data analysis. The handling of significant missing data is a challenge. In this paper, we are interested in the problem of 3D shape retrieval where the query shape is incomplete with moderate to significant portions of the original shape missing. The key idea of our method is to grasp the basis local descriptors for each shape in the retrieved database by sparse dictionary learning and apply them in sparsely coding the local descriptors of an incomplete query. First, we present a method of computing heat kernel signatures for incomplete shapes. Next, for each shape in the database, a set of basis local descriptors, which is called a dictionary, is learned and taken as its representative. Finally, a query incomplete shape’s heat kernel signatures are respectively reconstructed by each dictionary, and the shape similarities are therefore measured by the reconstruction errors. Experimental results show that the proposed method has achieved significant improvements over previous works on retrieving non-rigid incomplete shapes.