Brovey Transform Based Image Fusion For Impurities Segmentation And Detection On Edible Bird’s Nest
Edible bird’s nest (EBN) is one of the most important products in food and agricultural industry in South East Asia. In Malaysia, the production of EBN soaring because of the exportation of EBN to meet the demand of overseas market. Assurance of cleanliness is one of the major difficulties fac...
Saved in:
Main Author: | |
---|---|
Format: | Monograph |
Language: | English |
Published: |
Universiti Sains Malaysia
2019
|
Subjects: | |
Online Access: | http://eprints.usm.my/58276/1/Brovey%20Transform%20Based%20Image%20Fusion%20For%20Impurities%20Segmentation%20And%20Detection%20On%20Edible%20Bird%E2%80%99s%20Nest.pdf http://eprints.usm.my/58276/ |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Edible bird’s nest (EBN) is one of the most important products in food and agricultural
industry in South East Asia. In Malaysia, the production of EBN soaring because of the
exportation of EBN to meet the demand of overseas market. Assurance of cleanliness
is one of the major difficulties faced in processing the EBN. Current cleaning method
of EBN is labour dependency, time consuming and not cost effective. Automated
inspection was introduced but still continues to exist as a challenging field of
development as there is no effective algorithms for impurities segmentation. Some
impurities have similar colour as EBN features which increase the complexity of image
processing. In this study, Brovey transform based image fusion is used to highlight the
impurities in EBN and ease the segmentation process. Various types of Multispectral
(MS) reference images were considered in image fusion process. Comparison was made
to obtain the MS reference image with highest accuracy of segmented region. The
performances of fused images are evaluated based on segmentation rate, precision,
accuracy, error rate and dice similarity index (DSI). The optimal performances were
achieved by the green light without erosion MS reference image with an overall
segmentation rate of 49.96%, precision of 48.78%, accuracy of 40.00%, error rate of
60% and DSI of 0.571. |
---|