Segmentation and extraction of Songket motifs using region labeling / Mohd Sharizal Safaruddin

• Main Author: Safaruddin, Mohd Sharizal
• Format: Thesis
• Published: 2003
• Online Access: http://ir.uitm.edu.my/id/eprint/1688/
• Institution: Universiti Teknologi Mara

Songket motif is one of Malaysia traditional art that fill the richness of cultural heritage that reflected in the handicraft design products. Nowadays, a lot of abstract motifs exist that will eliminate the original motifs shape and structures. In order to collect the motifs, segmentation technique of an image is taken into this project. Segmentation is the intermediate level in image processing and it can be viewed as separating the image into disjoint or non overlapping regions. A study has made in this project by applying region or object labelling segmentation technique over the motifs. Before region labelling technique could be applied, the scanned photos of songket motifs have gone through preprocessing medium. The preprocessing task indicates three processes that are image conversion, image transformation and image enhancement. After segmentation process, extraction of motifs is performed to get the result of isolated motifs for each input images. Then, the measurement for accuracy of using region labelling technique is made from the observation of criteria that have been determined by from its correct segmented motifs result and actual existed motifs. Calculations for accuracy segmented motifs are using simple mathematical formulation in order to obtain its percentage performance. Eventually, it is shown that region labelling segmentation technique is useful to use in motifs segmentation and extraction. From the motifs result, a discussion is made due to some results that are shown is unexpected. Lastly, recommendation is pointed out to improve the technique of region labeling in segmentation of motifs.