Subramaniam Jayanti

Three-dimensional shape searching: state-of-the-art review and future trends

Three-dimensional shape searching is a problem of current interest in several different fields. Most techniques have been developed for a particular domain and reduce a shape into a simpler shape representation. The techniques developed for a particular domain will also find applications in other domains. We classify and compare various 3D shape searching techniques based on their shape representations. A brief description of each technique is provided followed by a detailed survey of the state-of-the-art. The paper concludes by identifying gaps in current shape search techniques and identifies directions for future research.

Developing an engineering shape benchmark for CAD models

Abstract Three-dimensional shape retrieval is a problem of current interest in several different fields, especially in the mechanical engineering domain. There exists a large body of work in developing representations for 3D shapes. However, there has been limited work done in developing domain-dependent benchmark databases for 3D shape searching. We propose a benchmark database for evaluating shape-based search methods relevant to the mechanical engineering domain. Twelve different shape descriptors belonging to three categories, namely: (1) feature vector-based, (2) histogram-based, and (3) view-based, are compared using the benchmark database. The main contributions of this paper are the development of a new engineering shape benchmark and an understanding of the effectiveness of different shape representations for classes of engineering parts. Overall, it was found that view-based representations yielded better retrieval results for a majority of shape classes, while no single method performed best for all shape categories.

Shape-based searching for product lifecycle applications

Estimates suggest that more than 75% of engineering design activity comprises reuse of previous design knowledge to address a new design problem. Reusing design knowledge has great potential to improve product quality, shorten lead time, and reduce cost. However, PLM systems, which address the issue of reuse by searching for keywords in filenames, part numbers or context attached to CAD models, do not provide a robust tool to search reusable knowledge. This paper presents a brief overview of a novel approach to search for 3D models. The system is built on a client-server-database architecture. The client takes in the query input from the user along with his search preferences and passes it to the server. The server converts the shape input into feature vectors and a unique skeletal graph representation. Details of the algorithms to perform these steps are presented here. Principal advantages of our graph representation are: (i) it preserves geometry and topology of the query model, (ii) it is considerably smaller than the B-Rep graph, and (iii) it is insensitive to minor perturbations in shape, but sensitive enough to capture the major features of a shape. The combined distance of feature vectors and skeletal graphs in the database provide an indirect measure of shape similarity between models. Critical database issues such as search system efficiency, semantic gap reduction and the subjectivity of the similarity definition are addressed. This paper reports our initial results in designing, implementing and running the shape search system.

A Reconfigurable 3D Engineering Shape Search System: Part I — Shape Representation

This paper presents an approach for a reconfigurable shape search system for 3D engineering models using a client-serverdatabase architecture. The current paper focuses on the server functionality, while a subsequent paper will focus on the database issues. The server takes the shape query as input from the client and converts it into feature vectors and a new skeletal graph representation which we have developed. The algorithms such as voxelization, skeletonization, and skeletal graph extraction for accomplishing these are described in detail. The principal advantages of the skeletal graph representation are that (i) it preserves the geometry and topology of the query model, (ii) it is considerably smaller than the B-Rep graph, and (iii) it is insensitive to minor perturbations in shape, while sensitive enough to capture the major features of a shape. Our representation is also synergistic with the human cognitive representation of shape. The results indicate that the skeletal graph is considerably smaller than the B-Rep graph even for complicated shapes.Copyright

A Reconfigurable 3D Engineering Shape Search System: Part II — Database Indexing, Retrieval, and Clustering

This paper introduces database and related techniques for a reconfigurable, intelligent 3D engineering shape search system, which retrieves similar 3D models based on their shape content. Feature vectors, which are numeric “fingerprints” of 3D models, and skeletal graphs, which are the “minimal representations of the shape content” of a 3D model, represent the shape content. The Euclidean distance of the feature vectors, as well as the distance between skeletal graphs, provides indirect measures of shape similarity between the 3D models. Critical database issues regarding 3D shape search systems are discussed: (a) database indexing, (b) semantic gap, (c) subjectivity of similarity, and (d) database clustering. An Rtree based multidimensional index is used to speed up the feature-vector based search operation, while a decision treebased approach is used for efficiently indexing/searching skeletal graphs. Interactions among users and the search system, such as relevance feedback and feature vector reconfiguration, are used to bridge the semantic gap and to customize the system for different users. Database clustering of the R-tree index is compared with that generated by a selforganizing map (SOM). Synthetic databases and real 3D model databases are employed to investigate the efficiency of the multidimensional index and the effectiveness of relevance feedback.Copyright

Navigation and Discovery in 3D CAD Repositories

Due to increasing product design complexities and the ever-expanding variety of product parts, the amount of information that designers must catalog has exploded. Accordingly, capable CAD tools to help designers create engineering artifacts are now pervasive. The volume of such engineering artifacts generated has increased exponentially and enterprises spend huge resources to organize and archive them into repositories. In these large design repositories, traditional text-based searches prove unwieldy and impractical, and are thus insufficient for individuals seeking 3D content. The paper explains that while traditional text-based searches are impractical for users seeking 3D content in large repositories, existing 3D search systems present search results in a 1D list, which is hard to search. A new interaction paradigm lets users navigate results in 2D and 3D spaces and easily find 3D models that are similar overall or in a single orientation.

Shape-based clustering for 3D CAD objects: A comparative study of effectiveness

3D shape retrieval and clustering is of current interest in several different fields, including mechanical engineering. Several new shape representations for 3D objects are continuing to emerge. Most shape representations are embedded in a variety of feature spaces. However, some of the recently reported shape representations are embedded in arbitrary metric spaces, i.e. distance spaces, rather than in multi-dimensional feature space. For such representations, the only operations available on the data objects are distance calculations between the objects. In addition, some of the view-based representations are embedded in non-metric spaces where the representations and the corresponding distances do not follow the triangle inequality. For shape clustering applications, most existing algorithms assume the shape representations either to be embedded in a multi-dimensional feature space or a metric distance space, making it difficult to evaluate several shape representations that do not conform to these assumptions. Therefore, two different approaches were evaluated for using the distance features of a shape to obtain clustering results. In the first method, the original distances are transformed into feature space using a multi-dimensional scaling approach for use with K-means clustering. The second approach directly uses the original distances with a distance-based clustering algorithm. We compared the clustering effectiveness of these two approaches using a classified benchmark database of 3D models. The effect of using different shape descriptors and number of clusters was studied using four measures of clustering effectiveness. Several statistical methods, including the Rand Index and Mutual Information Index, were used to objectively evaluate the clustering efficacy.

An Engineering Shape Benchmark for 3D Models

Three dimensional shape searching is a problem of current interest in several different fields, especially in the mechanical engineering domain. There has been a large body of work in developing representations for 3D shapes. However, there has been limited work done in developing domain dependent benchmark databases for 3D shape searching. In this paper, we propose a benchmark database for evaluating shape based search methods relevant to the mechanical engineering domain. Twelve feature vector based representations are compared using the benchmark database. The main contributions of this paper are development of an engineering shape benchmark and an understanding of the effectiveness of different shape representations for classes of engineering parts.Copyright

Effectiveness and efficiency of three-dimensional shape retrieval

The effectiveness and efficiency of a content-based three-dimensional shape search system are investigated for supporting the re-use of engineering designs. Search effectiveness is characterized by precision and recall values from repeated search experiments. We extract four feature vectors from each three-dimensional shape and compare them for effectiveness. The performance of an R-tree-based index structure is also evaluated to characterize search efficiency. Search efficiency is evaluated by the ratio of the number of visited nodes in a search operation to the number of nodes in a database index. A multi-step refinement approach is proposed to improve search effectiveness. Based on the results of the experiments with a database of real models, the effectiveness of using multi-step refinement is predicted to be 51% higher than that of one-shot search using a single feature vector, although the difference becomes smaller when the number of retrieved shapes is larger. The R-tree index significantly improves the efficiency of our search system. Search efficiency decreases with the dimensionality of the data records and the capacity of database nodes. Based on our experiments with the synthetic database, the efficiency is predicted to be stable when the size of a database increases.