Partial Matching on CAD Model Retrieval
Date Issued
2010
Date
2010
Author(s)
Tsai, Yi-Lung
Abstract
For engineering applications, a new design can be developed efficiently from an existing design with same functions, features, and manufacturing properties. Although several techniques have been developed for assessing the similarity among models, most methods are not sufficiently suitable for design and manufacturing. This work presents a graph-based retrieval architecture based on partial matching and a CAD model retrieval system based on the local feature correspondence. First construct attributed graph for each CAD model that retains the geometric and topological data from STEP format. Then transform the problem of model similarity assessment into the maximum clique problem and use simulated annealing algorithm to solve it. The proposed simulated annealing algorithm that successfully figures out several test files from DIMACS shows its practicability in similarity assessment. To achieve the goal of local feature correspondence, solve the graph-matching problem using independent maximal cliques (IMC) detection method. All independent maximal cliques, which represent common features between models, are identified using simulated annealing. According to IMC detection method, the experimental results demonstrate the practicability of this work for CAD model retrieval based on local feature correspondence.
Subjects
partial matching
similar CAD model retrieval
local feature correspondence
independent maximal cliques
simulated annealing
File(s)![Thumbnail Image]()
Loading...
Name
ntu-99-D94522019-1.pdf
Size
23.53 KB
Format
Adobe PDF
Checksum
(MD5):25b302615d6bdd9a065bacbd057e7836