Computer Vision/Computer Graphics Collaboration Techniques / Lecture Notes in Computer Science Bd.4418 (PDF)

Automatic Combination of Feature Descriptors for E.ective 3D Shape Retrieval (p.49)
Abstract.
We focus on improving the effectiveness of content-based 3D shape retrieval. Motivated by retrieval performance of several individual 3D model feature vectors, we propose a novel method, called prior knowledge based automatic weighted combination, to improve the retrieval effectiveness. The method dynamically determines the weighting scheme for different feature vectors based on the prior knowledge. The experimental results show that the proposed method provides significant improvements on retrieval e.ectiveness of 3D shape search with several measures on a standard 3D database. Compared with two existing combination methods, the prior knowledge weighted combination technique has gained better retrieval effectiveness.

1 Introduction

With the rapid development of 3D scanner technology, graphic hardware, and the World-Wide Web, there has been an explosion in the number of 3D models available on the Internet. In order to make use of these 3D models, the techniques of effective 3D shape retrieval become increasingly significant. 3D models can be annotated by keywords at first, facilitating the text-based retrieval. However, this is not a promising approach, because generally annotations are manually created, which is prohibitively expensive and subject to some factors.

To overcome the disadvantages of annotation-based approach, the so-called contentbased 3D shape retrieval, using the 3D model itself, has been proposed as an alternative mechanism [9]. In [17], Min compared four text annotation-based matching methods and four content-based retrieval approaches, and the experiments showed that the relatively simple solution of using only associated text for retrieval of 3D model was not as effective as using their shape.

As a promising approach applied in many fields, the content-based 3D shape retrieval