Intelligent garment system using fuzzy expert system / Amal Nabilah Md Amin
Body measurement is the most important thing that need to be considered in order to find the best fit size clothes. Traditionally, manual taking measurement by tailor has been applied. However, this conventional method requires more time and energy. This method also involved highly cost which may gi...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2007
|
| Online Access: | http://ir.uitm.edu.my/id/eprint/18254/ http://ir.uitm.edu.my/id/eprint/18254/2/TD_AMAL%20NABILAH%20MD%20AMIN%20CS%2007_5.pdf |
| Summary: | Body measurement is the most important thing that need to be considered in order to find the best fit size clothes. Traditionally, manual taking measurement by tailor has been applied. However, this conventional method requires more time and energy. This method also involved highly cost which may give a huge impact to the garment company. Therefore, a new system should be developed to measure human body based on the data (image) given by the customers. It will help to search and decide the best fit cloth(es) as a result. This research use one of the artificial intelligence techniques, Fuzzy Expert System. The type of fuzzy inference applied to the system is Mamdani with hedges. The first step applied is by defining the range of body measurement for each clothes size which was determined by the relevant expertise (tailor). Then fuzzy rules and fuzzy sets are determined. Next, coordinate point used for the image before the system gathers the measurement of the body. After that, the measurement will become an input to the knowledge base. The system will then be able to determine the match clothes size. This project was restricted to the Quppy’s Garment Sdn Bhd for the shoulder, chest, waist and hip range measurement. This project is significant to the garment industry whereby, in defining clothes sizes using intelligent systems may make process faster and decrease the cost. The system have been tested base on 30 images and 95% of the output generate are consistent. |
|---|