• [email protected]
  • +971 507 888 742
Submit Manuscript
SciAlert
  • Home
  • Journals
  • Information
    • For Authors
    • For Referees
    • For Librarian
    • For Societies
  • Contact
  1. Information Technology Journal
  2. Vol 8 (5), 2009
  3. 708-716
  • Issues
    Online First Current Issue All Issues
  • Information About
    Aims and Scope Editorial Board Guide to Authors Article Processing Charges
    Submit a Manuscript

Information Technology Journal

Year: 2009 | Volume: 8 | Issue: 5 | Page No.: 708-716
DOI: 10.3923/itj.2009.708.716
crossmark

Facebook Twitter Reddit Linkedin E-mail
Research Article

A New CAD Models Retrieval Method Based on Shape Similarity

Sun Wei
School of Mechanical Engineering, Dalian University of Technology, Dalian, 116024, China

Ma Tie-Qiang
School of Mechanical Engineering, Dalian University of Technology, Dalian, 116024, China

Guo Li
School of Mechanical Engineering, Dalian University of Technology, Dalian, 116024, China

ABSTRACT


The CAD model retrieval based on shape similarity is the research focus of computer graphics and computer-aided design field. To obtain higher retrieval precision and efficiency, a new CAD model retrieval method based on shape similarity is proposed. The method is divided into two steps. Firstly, breadth-first-search-based spanning tree algorithm is applied to obtain an initial boundary matching between retrieval object and retrieval condition. Secondly, the topology adjacency approximation algorithm is put forward to find the optimal boundary matching based on the initial boundary matching by the cycle and approximation process and the matching result is used to calculate the shape similarity between retrieval object and retrieval condition. In order to calculate the similarity between various types of boundary faces, a new similarity calculation method is present. Finally, a CAD model retrieval system on the platform of UG is developed based on the proposed method. Experimental results show that the proposed method is feasible and effective.
PDF Abstract XML References Citation

Keywords


  • retrieval
  • topology adjacency approximation
  • boundary matching
  • CAD
  • similarity calculation

How to cite this article

Sun Wei, Ma Tie-Qiang and Guo Li, 2009. A New CAD Models Retrieval Method Based on Shape Similarity. Information Technology Journal, 8: 708-716.

DOI: 10.3923/itj.2009.708.716

URL: https://scialert.net/abstract/?doi=itj.2009.708.716

INTRODUCTION


CAD model is one kind of most important engineering information and used to express the structure and shape of product. With the increase of CAD models, the reuse of CAD models becomes the key problem in product design. As the key technology of CAD model reuse, the CAD model retrieval based on shape similarity has been the research focus of CG and CAD field.

Content-based model retrieval method is the better solution to the problem. The so-called content-based retrieval method takes structures, shapes or other properties of model as the retrieval information and the specific descriptor is set to calculate the similarity between retrieval object and retrieval condition (Wan Li et al., 2007). Research on content-based retrieval method has been performed by many organizations or universities, such as Purdue University (Pu and Ramani, 2005; Lou et al., 2003; Hou and Ramani, 2007), Princeton University (Shilane et al., 2004), Carnegie-Mellon University (Zhang and Chen, 2001) and University of Maryland (Ip Cheuk and Gupta Satyandra, 2007). Most of them set focus on the retrieval method of mesh model and research on the retrieval method of CAD models are insufficient. Main methods of CAD model retrieval include the method based on statistics (Wang et al., 2006; El-Mehalawi and Miller Allen, 2006), the method based on graph-isomorphism algorithm (Liu and Hi, 2006; Rea et al., 2004; Pitta and Michael, 2005; Gal and Cohen-Or, 2007; Ou-Yang and Liu, 1999; El-Mehalawi and Miller Allen, 2006; Gao et al., 2006; Sundar et al., 2003; Hong et al., 2006; Bespalov et al., 2003) and the method based on boundary matching (Wang et al., 2007, 2008a, b; Ma et al., 2008) and the method based on boundary matching is considered as the best solution.

Wang et al. (2007, 2008a, b) applied the optimal matching algorithm of bipartite graph to find the optimal boundary matching between retrieval object and retrieval condition. The time complexity of the method is O(n3), since the time complexity of Kuhn-Munkres algorithm (Yin and Wu, 2003) is O(n3). Ma et al. (2008) used the shape-location code to describe the geometry and topology characteristics of boundary faces on CAD model and the shape-location codes of faces constitute the shape-location code matrix. The similarity between shape-location code matrices represents the similarity of CAD models. The average retrieval time of this method is about 0.4 sec. Wang et al. (2008c) built the attributed graphs of CAD models and applies sub-graph isomorphism algorithm to solve the problem on the partial structure retrieval of CAD models. The time complexity of the method is the interval (O(m),O(n!)/(n–m)!)). Where, n and m are the numbers of vertices in the big graph and the small graph, respectively.

In earlier mentioned methods, the boundary matching processes are all completed by one-step recycling. As known, boundary faces of CAD model are interdependent with topology adjacency relationships. In fact, the optimal boundary matching cannot be obtained by one-step recycling and the self-adaptive adjustment process is necessary during boundary matching process. Thus the retrieval precisions of these methods cannot be improved well.

In order to solve the problem, this study proposes the topology adjacency approximation algorithm and applies the algorithm to adjust the boundary matching process between retrieval object and retrieval condition self-adaptively. The relatively accurate similarity can be gotten by the algorithm. Moreover, this study proposes a uniform method to calculate the similarity between various types of faces. Finally, a retrieval system on UG platform is developed according to the proposed method and the retrieval effect of the proposed method is verified by experiments.

TOPOLOGY ADJACENCY APPROXIMATION

The similarity calculation between retrieval object and retrieval condition is the basis of CAD model retrieval. The differences lie in the similarity calculation method between retrieval object and retrieval condition for different retrieval methods. This study proposes the topology adjacency approximation algorithm and applies it in the similarity calculation of CAD models. The framework of the method is shown in Fig. 1.

Figure 1 shows that the method divided into two steps. The first step is the initial boundary matching process and the second step is the topology adjacency approximation process. The first step provides the initial boundary matching for the second step and an appropriate initial boundary matching will accelerate the convergence of the method. The initial boundary matching is obtained by comparing the geometrical similarities of faces between retrieval object and retrieval condition. Thus the geometry similarity calculation of faces is the key problem during initial boundary matching. The topology adjacency approximation algorithm is a cycle and approximation process. For each cycle, the boundary matching between retrieval object and retrieval condition will be adjusted.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 1: Framework of the proposed method

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 2: The topology adjacency approximation process

The matching relationships of boundary faces will be exchanged for arbitrary two couples of matched faces if the exchange will increase the similarity value between retrieval object and retrieval condition. With each cycle process, the topology similarity of boundary faces will be recalculated between retrieval object and retrieval condition, since the boundary matching is changing constantly with the topology adjacency approximation process. Figure 2 shows the topology adjacency approximation process during searching for the optimal boundary matching.

SIMILARITY CALCULATION OF BOUNDARY FACES

Geometry similarity calculation: For two boundary faces f and f’, if the shape of f is similar to that of f’ in vision sensation, f and f’ are considered as a couple of geometry similar boundary. However, other methods cannot calculate the geometry similarity between arbitrary two boundary faces if the types of them are different. This study proposes a uniform method to calculate the geometry similarity between different types of faces.

Curvature similarity calculation between boundary faces: Figure 3 shows an boundary face f and the centroid of f is denoted as Wf. On f, P is the closest point from Wf. The normal vector of f is tagged as Image for - A New CAD Models Retrieval Method Based on Shape Similarityat point P. If the vector from P to Wf is expressed as Image for - A New CAD Models Retrieval Method Based on Shape Similarity, the dot product between Image for - A New CAD Models Retrieval Method Based on Shape Similarityand Image for - A New CAD Models Retrieval Method Based on Shape Similaritywill be Image for - A New CAD Models Retrieval Method Based on Shape Similarity.

For a couple of matched boundary faces f and f’, the curvature description information of both them is as shown in Fig. 3. The curvature similarity Sc(f, f’) between f and f’ can be calculated by Eq. 1.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(1)

In Eq. 1, Sm(f, f’) and Sa(f, f’) are the curvature similarities between f and f’ in diretion of maximal curvature and direction of minimum curvature, respectively. The direction of maximal curvature is the tangent direction of boundary face at certain point, in which the absolute value of curvature reaches the maximal. Correspondingly, the direction of minimum curvature is the tangent direction of face at certain point, in which the absolute value of curvature reaches the minimum. Sm(f, f’) and Sa(f, f’) can be gotten by function g(x) and g(x) is as follows:

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(2)

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 3: Curvature description of boundary face

While Eq. 2 is applied to calculate Sm(f, f’), Image for - A New CAD Models Retrieval Method Based on Shape Similarity. While Eq. 2 is used to calculate Sa(f, f’), Image for - A New CAD Models Retrieval Method Based on Shape Similarity. Where, k and k’ are the curvatures of f and f’ at the centroids of them separately. η is defined as a threshold which limits the comparable scope between the curvatures of f and f’ in directions of maximal curvature and minimum curvature. If x is less than η, the curvature similarity between f and f’ will be zero. The curvature similarity describes the curvature similarity, the curvature direction similarity and the area distribution similarity between boundary faces.

Shape-position similarity calculation between boundary faces: For arbitrary boundary face, the shape of it can be represented by the external loop of it. This study codes the binary code of external loop to describe the shape and position of boundary face. For a boundary face f, the external loop of f can be denoted as directed edges set LExt(f) = {e1, e2,…,en}, in which any edge is connected to the next edge of it in counter clockwise. The binary code of f is denoted as SP(f) and SP(f) is gotten by Eq. 3.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(3)

The function BIN is the code function. The binary code of convex edge will be 1 and that of concave edge is 0. For CAD model in Fig. 4, the external loop of f is express as directed edges set LExt(f) = {e1, e2, e2, e4}. The concavity-convexity code of f is SP(f) = 1010. Obviously, the external loop of boundary face is divided into several sections and the concavity-convexity characteristic of all directed edges in same section is same. Such a section is called concavity-like section or convexity-like section if all edges in the section are convex or concave.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 4: Calculation of shape-position similarity

Both concavity-like section and convexity-like section are called direction-like section uniformly.

Figure 4 shows an example for calculation of shape-position similarity between boundary faces. It is hypothesized that the numbers of directed edges on the external loops of f and f’ are n, respectively. t and t’ are the numbers of direction-like sections on the external loops of f and f’, respectively. The following relationship will be satisfied:

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(4)

In Eq. 4, mi and ni are the number of directed edges on the ith direction-like section on the external loop of fand that of f’ individually. If t>t’, the shape-position similarity between f and f’ is denoted as Sp(f, f’) and Sp(f, f’) is computed by Eq. 5.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(5)

It is demanded that the binary code of f’ should be compared with that of f circularly, since each directed edge is connected to the next edge of it circularly on the external loop of boundary face. Sp(f, f’) should take the maximal element in set {Sp,j(f, f’)|1≤j≤t–1}. Sp,j(f, f’) is the shape-position similarity obtained bythe jth comparison between f and f’. Equation 6 is used to calculate Sp,j(f, f’).

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(6)

In Eq. 6, λ is a Boolean variant. λ = 1, if the type of direction-like section on the external loop of f’ is same to the type of its corresponding section on the external loop of f, otherwise λ = 0.

Geometry similarity between boundary faces: Sc(f, f’) and Sp(f, f’) constitute the geometry similarity SG(f, f’) between f and f’ and Eq. 7 is applied to calculated SG(f, f’).

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(7)

For arbitrary couple of matched faces, the geometry similarity between them is gotten by Eq. 1-7, though their types may be different.

Topology similarity calculation: It is hypothesized that M and M’ are different CAD models. M is taken as the retrieval condition and M’ is the retrieval object. The set of boundary faces on model M is denoted as FS and FS = {f1, f2,…,fn}. Where, n is the number of boundary faces on model M. For arbitrary boundary face fi(1≤I≤n) on M, f1’ is the matching face of fion M’. The number of adjacent boundary faces of fion hi and the number of adjacent boundary faces of fi is hi. The adjacent boundary faces of fi are fi,1, fi,2,…,fihi and the matching faces of them are Image for - A New CAD Models Retrieval Method Based on Shape Similarityon M’, respectively. Figure 5 shows the mapping relationship of boundary faces from M to M’.

On the topology adjacency graph of M’, the distances from fi’ to Image for - A New CAD Models Retrieval Method Based on Shape Similarityare defined as d(fi’, fi,j’) (1≤j≤hi), respectively. The average value of d(fi’, fi,j’) (1≤j≤hi), is expressed as Image for - A New CAD Models Retrieval Method Based on Shape Similarityand Eq. 8 is applied to calculate Image for - A New CAD Models Retrieval Method Based on Shape Similarity.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(8)

The topology similarity between fi’ and fi is denoted as ST(fi, fi’). The value of ST(fi, f’i) is obtained by Eq. 9.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(9)

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 5: Mapping relationship of boundary faces from M to M’

In Eq. 9, ζ is named topology factor. Equation 9 shows that the topology similarity will increase with decrease of average distance Image for - A New CAD Models Retrieval Method Based on Shape Similarity.

SIMILARITY CALCULATION OF CAD MODELS

Initial boundary matching: In order to calculate the boundary matching between retrieval object and retrieval condition, Wang et al. (2007, 2008a, b) applied K-M algorithm to find out the boundary matching. Because, the time complexity of K-M algorithm is O(n3) and all matching faces cannot converge to some partial structure, the efficiency and accuracy of this CAD model retrieval is lower. Aiming at solving these problems, this study proposes a boundary matching method based on the spanning tree algorithm. Though, the matching result of this method cannot be used to calculate the exact similarity between CAD models, it can be taken as the input of the topology adjacency approximation algorithm. The tree is spanned from the topology adjacency graph of retrieval condition. All matching faces of retrieval condition are obtained with the spanning process of the tree. It is hypothesized that M1 and M2 are retrieval object and retrieval condition, respectively. The set of boundary faces on M1 is FS1 and FS1 = {f1,1, f1,2,…f1,n}. n is the number of boundary faces on M1. The set of boundary faces on M2 is FS1 and FS2 = {f2,1, f1,2,…f2,m}. M2 is the number of boundary faces on M2. The method is described as follows:

Geometry similarity calculation between boundary faces: f1,i(1≤I≤n) and f2,j(1≤I≤n) are boundary faces on M1 and M2, respectively. The geometry similarity between f1,i and f2,i is denoted as SG(f1,i, f2,j). SG(f1,i, f2,j) is calculated by Eq. 1-7. The geometry similarity matrix is expressed as Anxm.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity

Seeking the root node of spanning tree: SG(f1,i, f2,j) (f’1,i∈ FS2, 1≤I≤n) is located on each row of Anxm and f’1,i, is the matching boundary face of f1,i. The following relationship should be satisfied.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity

If f1,i,k (1≤k≤hi, f1,i,k ∈ FS1) is the adjacent faces of f1,i, SG(f1,i,k, f’1,i,k) (f’1,i,k ∈ FS2) should be gotten and the following relationship should be satisfied.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity

The average distances Image for - A New CAD Models Retrieval Method Based on Shape Similarityfrom f’1,i to f’1,i,k (1≤k≤hi) is calculated by Eq. 8. Finally, a couple of matched faces f1,r and f’1,r(f1,r, FS1, f’i,r ∈ FS2) should be obtained and Image for - A New CAD Models Retrieval Method Based on Shape Similarity. f1,r will be the root of spanning tree.

Initial boundary matching based on spanning tree algorithm: f1,r is taken as the root node of the spanning tree. The spanning tree is generated by using Breadth First Search (BFS) algorithm to traverse the topology adjacent graph of M1. With the growth of the spanning tree, it is necessary to search for the matching face of newly generated tree node on M2. If f1,i(1≤I≤n) is a tree node of the spanning tree and the matching face of f1,i is denoted as f’i,i (f’i,i ∈ FS2), f’i,i(1≤I≤n) constitutes the set FS1’ = {f’i,r|1≤I≤n} and FS1’⊆FS2.

Initial similarity calculation between CAD models: If the similarity between M1 and M2 is denoted as S(M1, M2), will be expressed as the following equation.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(10)

In Eq. 10, ST(f1,i, f’1,i) is the topology similarity between f1,i and f’1,i(1≤I≤n). wi is a value which represents the weight of f1,i in the similarity between M1 and M2. Equation 11 is applied to calculate wi.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
(11)

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 6: Initial boundary matching

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 7: Topology adjacency approximation algorithm

Figure 6 shows the process of initial boundary matching. The time complexity of initial boundary matching process is O(n2), since the time complexity of BFS-based spanning tree algorithm is O(n2). The advantages of above-mentioned method is that the initial boundary matching process can accelerate the convergence of topology adjacency approximation algorithm.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 8: Process of topology adjacency approximation

Topology adjacency approximation algorithm: The topology adjacency approximation algorithm is proposed to search for the optimal boundary matching between retrieval object and retrieval condition and make the boundary matching converge to the global optimal solution. The global optimal solution should meet the requirement of the absolute error. It is hypothesized that the initial similarity between M1 and M2 is denoted as S0. Figure 7 shows the topology adjacency approximation algorithm.

Figure 8 also shows the detailed process of topology adjacency approximation. The kernel process lies in that the matching relationship between arbitrary two couple matched faces should be exchanged if the exchange can make the similarity increase. In fact, the topology adjacency approximation algorithm makes the boundary matching reach a stable status. In the status, the similarity between M1 and M2 will converge to an accurate value relatively.

Furthermore, the topology adjacency approximation algorithm is proven convergent by more experiments and its total cycle number is about γxnxm. Further experiments show that γ≤5. Therefore, the time complexity of the algorithm is O(n2). Moreover, the time complexity of initial boundary matching is also O(n2). Thus, the total time complexity of the proposed CAD model retrieval method will be O(n2).

APPLICATION IN CAD MODEL RETRIEVAL

Function and implementation: A CAD model retrieval system based on CAD platform is developed according to the above-mentioned method. The retrieval system takes CAD model or structures of CAD model as the retrieval condition. The system is divided into two modules. The first module is used to extract the retrieval condition from CAD model and the second is the retrieval engine used to search for the retrieval results by calculating and ordering the similarities between retrieval objects and retrieval condition. The functions of two modules will be analyzed in the following:

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 9: The interface of condition extraction module

Condition extraction module: The geometry information is extracted from CAD model on level of feature by condition extraction module. For example, F is a feature of CAD model and the set of boundary faces on F can be denoted as F = {f1, f2,…,fn}. n is the number of boundary faces on F. For any structure S on CAD model, the boundary information of S should be expressed as S = F1∪F2∪…∪Fm and m is the number of features on the structure S. Therefore, the condition extraction process is the extraction process of features F1, F2,..,Fm. Moreover, all boundary faces of the extracted structure should be keep interconnectivity and all the boundary faces constitutes a connected graph. Thus a connectivity detection algorithm should be necessary. This study uses the Breadth-First-Search (BFS) algorithm to check the connectivity of graph composed of all boundary faces on the structure. This study developed the condition extraction module on UG platform by VC++6.0 and the internal application programming interface of UG system. The interface of the condition extraction module is shown in Fig. 9.

Retrieval engine module: The similarity between retrieval object and retrieval condition is calculated by retrieval engine module. The similarity shows whether there is the structure being similar to the retrieval condition on the retrieval object. Thus, retrieval engine module is the main module used to realize the retrieval function. Retrieval engine module includes two main functions. The first function is to extract the geometry information from retrieval object and the second function is to calculate the similarity between retrieval object and retrieval condition. Whereas, the extraction function of geometry information of retrieval engine module is different from that of the condition extraction module. The retrieval engine module directly extracts all geometry information of retrieval object by traversing all boundary faces of retrieval object.

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 10: The interface of retrieval engine

All boundary faces of retrieval object constitute a connected graph. The retrieval engine module calculates the similarities between every retrieval objects and retrieval condition one by one. The model retrieval is completed by sorting the similarities between all retrieval objects and the retrieval condition and the retrieval results meeting the retrieval threshold will be returned. The retrieval threshold is input by users. This study developed the retrieval engine module on UG platform by VC++6.0 and the external application programming interface of UG platform. The interface of the retrieval engine module is shown in Fig. 10.

Retrieval precision: Precision and recall are the most important indexes to evaluate the retrieval method. As pointed out by Shilane et al. (2004), a precision-recall plot describes the relationship between precision and recall in a ranked list of matches. For each retrieval condition model in class C and any number K of top matches, recall (the horizontal axis) represents the ratio of models in class C returned within the top K matches, while precision (the vertical axis) indicates the ratio of the top K matches that are members of class C. A perfect retrieval result produces a horizontal line across the top of the plot (at precision = 1.0), indicating that all the models within the retrieval object’s class are returned as the top ranked matches. Otherwise, curves that appear shifted up represent superior retrieval results. In order to describe the precision curves of the model retrieval system, 867 UG models are built as retrieval samples with reference to the models’ shapes in the Engineering Shape Benchmark (ESB) of Purdue University (http://shapelab.ecn.purdue.edu/Benchmark.aspx) and a large number of experiments are conducted.

Table 1: Retrieval results
Image for - A New CAD Models Retrieval Method Based on Shape Similarity

Image for - A New CAD Models Retrieval Method Based on Shape Similarity
Fig. 11: Precision-recall curves

The precision-recall plot is drawn by analyzing the experimental results. Figure 11 is the precision-recall plot. The curve I-IV represent the precision-recall curves which are obtained, respectively by the retrieval methods proposed by Wang et al. (2006, 2007) and Ma et al. (2008) and present study.

Figure 11 shows that the precision of the method is higher than those of other methods in same condition of recall ratio. Especially the method has obvious superiority to other methods in recall interval (0, 0.5).

EXPERIMENTS

Many retrieval experiments have been conducted. In order to demonstrate the retrieval effect of the proposed method, two CAD models in the 1st column of Table 1 are taken as the retrieval conditions and ζ, ε and the retrieval threshold are taken as 0.50, 0.0001 and 0.6, respectively. The retrieval results are returned and shown in the 2nd column of Table 1.

CONCLUSIONS


In summary, this study investigates the CAD model retrieval based on boundary matching. A new retrieval method based on topology adjacency approximation algorithm is proposed. The proposed retrieval method is applied to develop the CAD model retrieval system based on UG platform. Experimental results show that the proposed method is feasible. The performances of the model retrieval system are as follows:

• The adaptability of the proposed method is better than other method. It is not only used in the CAD model retrieval but also applied in the partial structure retrieval of CAD model
• The model retrieval is higher efficiency and its time complexity is about O(n2)
• The precision-recall plots shows that the precision of the proposed method is higher

ACKNOWLEDGMENTS


The initial funding of this research comes from the Program for New Century Excellent Talents in University, Ministry of Education, China (Grant No. NCET-05-0285). Supplemental support from the Science and Technology Key Projects of Liaoning Province in China is acknowledged (Grant No. 2006219012).

REFERENCES


  1. Wan Li, L., Q. Zhao and A. Hao, 2007. A method of 3D model retrieval by the spatial distributions of components. J. Softw., 18: 2902-2913.
    CrossRef

  2. Pu, J. and K. Ramani, 2005. A 3D model retrieval method using 2D freehand sketches. Lect. Notes Comput. Sci., 3515: 343-346.
    Direct Link

  3. Lou, K., S. Jayanti and N. Iyer, 2003. A reconfigurable 3D engineering shape search system part II: Database indexing, retrieval and clustering. Proceedings of the ASME Design Engineering Technical Conferences and Computers and Information in Engineering Conference, September 2-6, 2003, Chicago, IL., USA., pp: 169-178.
    Direct Link

  4. Hou, S. and K. Ramani, 2007. Classifier combination for sketch-based 3D part retrieval. Comput. Graph., 31: 598-609.
    CrossRef

  5. Shilane, P., P. Min, M. Kazhdan and T. Funkhouser, 2004. The princeton shape benchmark. Proceedings of the Shape Modeling International, (SMI'04), Los Alamitos, pp: 167-178.
    Direct Link

  6. Zhang, C. and T. Chen, 2001. Indexing and retrieval of 3D models aided by active learning. Proceedings of the 9th ACM international conference on Multimedia, September 30-October 5, 2001, ACM New York, USA., pp: 615-616.
    CrossRefDirect Link

  7. Ip Cheuk, Y. and K. Gupta Satyandra, 2007. Retrieving matching CAD models by using partial 3D point clouds. Comput. Aided Des. Applied, 4: 629-638.
    Direct Link

  8. Wang, F., S. Zhang and X. Bai, 2008. 3D model retrieval based on both the topology and shape features. J. Comput. Aided Des. Comput. Graph., 20: 99-103.

  9. Wang, Y., H.J. Ma and W. He, 2006. Clustering and retrieval of mechanical 3D CAD models. Comput. Integr. Manuf. Syst., 12: 924-928.
    Direct Link

  10. El-Mehalawi, M. and R. Miller Allen, 2006. A database system of mechanical components based on geometric and topological similarity, part: Indexing, retrieval, matching and similarity assessment. Comput. Aided Des., 35: 95-105.
    CrossRef

  11. Liu, W. and Y.J. He, 2006. Retrieving 3D CAD model with regional entropy distributions ICAT 2006. Lect. Notes Comput. Sci., 4282: 871-879.
    Direct Link

  12. Rea, H.J., J.R. Corney, D.E.R. Clark and N.K. Taylor, 2004. A surface partitioning spectrum (SPS) for retrieval and indexing of 3D CAD models. Proceedings of the 2nd International Symposium on 3D Data Processing, Visualization and Transmission, September 6-9, 2004, IEEE Computer Society Washington, DC, USA., pp: 167-174.
    CrossRefDirect Link

  13. Pitta, C. and M.M. Michael, 2005. Intelligent retrieval and reuse of CAD solid models. Proceedings of the International Conference on Robotics and Automation, April 18-22, 2005, IEEE Computer Society, USA., pp: 1413-1418.
    Direct Link

  14. Gal, R. and D. Cohen-Or, 2007. Salient geometric features for partial shape matching and similarity. http://www.cs.princeton.edu/~funk/sgp06.pdf.

  15. Ou-Yang, C. and P.Y. Liu, 1999. Applying the topological relationships of form features to retrieve part files from a CAD system. IIE Trans., 31: 323-337.

  16. Gao, W., S.M. Gao, Y.S. Liu, J. Bai and B.K. Hu, 2006. Multiresolutional similarity assessment and retrieval of solid models based on DBMS. Comput. Aided Des., 38: 985-1001.
    Direct Link

  17. Sundar, H., D. Silver, N. Gagvani and S. Dickinson, 2003. Skeleton based shape matching and retrieval. Proceedings of the Shape Modeling International, May 12-15, 2003, IEEE Computer Society Washington, DC, USA., pp: 130-139.
    CrossRefDirect Link

  18. Hong, T., K. Lee and S. Kim, 2006. Similarity comparison of mechanical parts to reuse existing designs. Comput. Aided Des., 38: 973-984.
    Direct Link

  19. Bespalov, D., W.C. Regli and A. Shokoufandeh, 2003. Reeb graph based shape retrieval for CAD. Proceedings of the ASME Design Engineering Technical Conference, September 2-6, 2003, Chicago, IL, USA., pp: 229-238.
    Direct Link

  20. Wang, H.S., S.S. Zhang and X. Bai, 2007. Optimal-matching-based 3D CAD model similarity assessment algorithm. Comput. Integr. Manuf. Syst., 17: 1921-1927.
    Direct Link

  21. Wang, H., S. Zhang and X. Bai, 2008. A partial retrieval algorithm of 3D CAD models based on attributed graphs. J. Comput. Aided Des. Comput. Graph., 20: 316-320.
    Direct Link

  22. Ma, L., Z. Huang and Q. Wu, 2008. Retrieval of CAD models based on shape-location codes of faces. J. Comput. Aided Des. Comput. Graph., 20: 19-25.
    Direct Link

  23. Wang, F., S. Zhang and X. Bai, 2008. Local matching of 3D CAD models based on subgraph isomorphism. J. Comput. Aided Des. Comput. Graph., 20: 1078-1084.
    Direct Link

  24. Yin, J.H. and K.Y. Wu, 2003. Graph Theory and Its Algorithm. University of Science and Technology of China Press, Hefei.

Leave a Comment


Your email address will not be published. Required fields are marked *

Useful Links

  • Journals
  • For Authors
  • For Referees
  • For Librarian
  • For Socities

Contact Us

Office Number 1128,
Tamani Arts Building,
Business Bay,
Deira, Dubai, UAE

Phone: +971 507 888 742
Email: [email protected]

About Science Alert

Science Alert is a technology platform and service provider for scholarly publishers, helping them to publish and distribute their content online. We provide a range of services, including hosting, design, and digital marketing, as well as analytics and other tools to help publishers understand their audience and optimize their content. Science Alert works with a wide variety of publishers, including academic societies, universities, and commercial publishers.

Follow Us
© Copyright Science Alert. All Rights Reserved