| Reusing engineering data has opened a new opportunity to improve product quality, shorten design time and reduce costs using know-how in the design process. Any manufacturer has an accumulated know-how related to design, production and performance of existing and previously manufactured products. Accessibility and possibility of reusing this accumulated knowledge is a key factor for optimizing design and performance of a new product. A potential algorithm which clarifies similarity between a new possible product and existed objects in the product data base enables designer to easily find a base for design of the new product. So the new design can be well optimized using know-how from design, production and performance of an existing product. Shape similarity assessment is a fundamental geometric reasoning problem for any comprehensive similarity comparison in product design applications, and has been subject of various researches beside other aspects. In this regard, Institute of Computer Aided Design earlier had a research over methods in the categories of Statistic-based, 3D Object Recognition-based and Feature Recognition-based. To complete previous study, the current diploma thesis work focuses on four following categories in order to have an inclusive evaluation of all methods used to convert a shape into a signature for comparison.
Aim and objectives:
- Study and evaluation of geometrical similarity comparison methods based on four categories of: Invariant/descriptor-based, Harmonics-based, Graph-based, Group Technology
- Review all relevant methods of mentioned categories with related case studies in the open literature
- Propose the best method of each category respect to advantages and disadvantages, with the list of predefined applications
- Implement the best method of each category for a predefined application in MATLAB
- Evaluate and test the performance of each method using MATLAB codes
Requirements:
- Familiarity with similarity comparison concept
- High potential in reading and analyzing the related documents
- Programming in MATLAB
|
UZ) | © Universität Stuttgart | Impressum