A search engine for 3D models

Abstract

As the number of 3D models available on the Web grows, there is an increasing need for a search engine to help people find them. Unfortunately, traditional text-based search techniques are not always effective for 3D data. In this article, we investigate new shape-based search methods. The key challenges are to develop query methods simple enough for novice users and matching algorithms robust enough to work for arbitrary polygonal models. We present a Web-based search engine system that supports queries based on 3D sketches, 2D sketches, 3D models, and/or text keywords. For the shape-based queries, we have developed a new matching algorithm that uses spherical harmonics to compute discriminating similarity measures without requiring repair of model degeneracies or alignment of orientations. It provides 46 to 245% better performance than related shape-matching methods during precision--recall experiments, and it is fast enough to return query results from a repository of 20,000 models in under a second. The net result is a growing interactive index of 3D models available on the Web (i.e., a Google for 3D models).

Shape-similarity search of 3D models by using enhanced shape functions

“…While AD and AAD have computational cost somewhat higher (about 1.5 times) than the D2, they significantly outperformed D2 in our evaluation experiment. While the AD and AAD might have the performance lower than the more elaborate shape features, such as the spherical harmonics based feature used in [Funkhouser03]. However, the computational costs of AD and AD2 estimated to be significantly lower than that of the sophisticated shape feature used in [Funkhouser03].…”

“…It is also extremely difficult to describe by words shapes that are not in the well-known shape or semantic categories. It is thus necessary to have a content-based search and retrieval systems for 3D models that are based on the features intrinsic to the 3D models, most important of which is the shape [Paquet97, Suzuki98, Keim99, Elad00, Paquet00, Regli00, Suzuki00, McWherter01, Osada01, Novotni01, Hilaga01, Veltkamp01, Vranic01, Zahaira01, Corney02, Ibato02, Mukai02, Ohbuchi02, Osada02, Zahaira02, Funkhouser03,Tangelder03].…”

From Low-Cost Depth Sensors to CAD: Cross-Domain 3D Shape Retrieval via Regression Tree Fields

“…Due to the simplicity, various techniques have been developed to retrieve 3D models whose 2D images match the query sketch. For instance, Funkhouser et al used a variant of the 3D sphere harmonics to develop a shape search engine that accepts sketches as queries [2]. Yoon et al employed suggestive contours and diffusion tensor fields to improve the robustness against shape and pose variance that often occurs in the user sketched images [17].…”

“…In the past two decades, extensive efforts have been made to design effective 3D shape retrieval algorithms [1]. The existing work is mainly focused on two search scenarios, i.e., search by sketch [2][3] (Figure 1(a)) and search with CAD models as query input [1] (Figure 1(b)). Along with the advances of low-cost depth sensors such as Microsoft Kinect, PrimeSense sensors, and the newly revealed mobile depth sensor from Google [4], there is tremendous growth of user-generated 3D data, which promotes the study of a new cross-domain retrieval problem, i.e., search with user-captured models, where the users capture potentially noisy depth data and images of the object to their interest, and then use reconstructed 3D models as queries to find similar 3D shapes from a large collection of high-quality CAD models as illustrated in Figure 1(c).…”

Autotagging to improve text search for 3D models

“…Existing 3D search engines require users to submit complex queries, such as drawing a sketch [1,4] or providing an initial 3D model as a starting point. In contrast, the simplest and most natural interface, keyword search on the descriptive text associated with each model, is limited by a dependence on the accuracy of these descriptions, and more fundamentally by the requirement that they even exist.…”