As the Queen of Textiles, silk and its related products have been very popular since ancient time. Silk fabric has been widely accepted because of its specialty in terms of high durability, light weight, and natural shine. Particularly, the uniqueness of the silk fabric comes from special skill and experience of handicraftsmen.
In silk manufacturing as well as in the preservation of good silkworm breeders, one key success factor depends on the butterflies (Bombyx mori) on how they can offer fecundity, fertilization, and high hatching rate. In Thailand, the Department of Sericulture assists in preparing and providing silkworm eggs laid out in a sheet of paper to farmers. As it is impossible to count thousands of these small eggs (1 mm in diameter) via naked eyes, the quantity of silkworm eggs is indirectly measured through volume and weight analysis, leading to large variation in the number of eggs and unfairness for farmers.
This manual inspection also implies that it is impractical for large scale production in supporting sericulture industry, community enterprises, and handicraft people.
To efficiently determine the number of silkworm eggs in each paper sheet, a simple machine vision system is employed. In our work, an off-the-shelf desktop scanner shines a beam of white light on a paper sheet that contains thousands of silkworm eggs. The reflected light from those thousand silkworm eggs creates a visible image. It is then analyzed by our image processing operations that include morphological and statistical learning processes. We name our system “Smart KhaiMai” (“KhaiMai” means silkworm egg in Thai language).
As a result, the locations of normal and defected eggs are automatically determined. Location of silkworm eggs leads to the number of silkworm eggs. There are also three incubation periods of silkworm eggs, namely, fecundity, fertilization, and hatching-out periods that can be pinpointed under our Smart KhaiMai V.2.5 as shown in figure above. For each period of incubation, normal and defected eggs can be effectively identified and are marked by red and blue dots, respectively, see figure below.
As shown in Figure below, our system has simple user interface with all silkworm eggs are highlighted. The fastest processing time is 1,000 eggs/second while the average speed is 200 eggs/second. The speed of the system is much faster than typical manual counting in the range of 120 and 600 times. The performance is varied between 70-95% of accuracy. These important system issues depend on the background color of the selected paper sheet and the process of laying out silkworm eggs on the paper.
Currently, our Smart KhaiMai V.2.5 has been installed and in utilization in 20 centers of Queen Sirikit Sericulture Centers in Thailand.
1 – Kiratiratanapruk, K.; Sinthupinyo, W., “Worm egg segmentation based centroid detection in low contrast image,” International Symposium on Communications and Information Technologies (ISCIT), pp. 1139-1143, Gold Coast, Australia, 2012.
2 – Kiratiratanapruk, K.; Methasate, I.; Watcharapinchai, N.; Sinthupinyo, W., “Silkworm eggs detection and classification using image analysis,” International Computer Science and Engineering Conference (ICSEC), Khon Kaen, Thailand, pp. 340-345, 2014.
Kantip Kiratiratanapruk received her B.Eng in Electrical Engineering from Kasetsart University, Bangkok, in 1997. She completed M.Eng in Electrical Engineering from Chulalongkorn University in 2001. She has been a researcher at Thailand’s National Electronic and Computer Technology Center (NECTEC) since 2002. Her research interest includes digital image processing and computer vision.
Ithipan Methasate received his B.Eng. and M.Eng. in Electrical Engineering from Chulalongkorn University and Ph.D. in Information Technology from Thammasat University, Thailand, in 1997, 2000, and 2010 respectively. He is a researcher at Thailand’s National Electronics and Computer Technology Center (NECTEC). His current research interests include image processing, video processing, language processing, and machine learning.
Wasin Sinthupinyo received his B.Sc. (Mathematics) from Khonkaen University, Khonkaen in 1991. He went on to complete his M.Sc. in Computer Science from Chulalongkorn University in 1995. He has been a researcher at Thailand’s National Electronic and Computer Technology Center (NECTEC) since 1995. His research interest includes digital image processing and computer vision.