A semantic recognition and rule-oriented approach for developing a product design is proposed. The development of an office chair is used to describe the design procedures. A number of referential products are used to quantify partial correlation coefficients, which represent individual contributions of constituent components to the impression of a whole product with respect to an image, by using the 'quantification theory type I' and 'semantic differential method'. The results are then built as an image database. A computer program is constructed by integrating the techniques of constructing B-splines, shape grammars and fuzzy operations to provide a CAD environment in which 3D basic models of a product can be constructed by starting with inputting an image word(s) and the corresponding membership functions for describing the image of the required product. After a basic product model has been constructed, new product forms can further be generated by regulating the shapes of the component(s) using shape rule(s).
All Science Journal Classification (ASJC) codes
- Arts and Humanities (miscellaneous)
- Social Sciences(all)
- Computer Science Applications
- Artificial Intelligence