ABSTRACT:
The need for pattern recognition system cannot be over emphasized as it cuts across many fields. Majority of the work on fabric pattern recognition focuses on determining the nature of the wefts and wraps in a given cloth. Others are more concerned with defect detection of hand woven fabrics and a few consider recognizing some African fabrics. However, there are limited studies on Saki Pattern recognition. Saki is a traditional Fulani hand woven material worn as everyday cloth by the Fulani Clan of West and Central Africa. This research aims to recognize Saki Pattern in woven fabrics using neuro-fuzzy system. A total of 600 images from four (4) different samples of Saki are collected and pre-processed to extract relevant features. The images are trained using Backpropagation algorithm (BP) Neural Network in Matlab environment. Fuzzy inference rules are then used for classification. The experimental results obtained showed that all four (4) Saki samples were predicted accurately with an average of 80% similarity. Thus, providing a lot of information on Saki which will help preserve the Fulani cultural heritage and boost the Saki textile industry.
Keywords:
Fuzzy inference, Image processing, Neural network, Pattern recognition, Saki
