Image Surface Defects Extraction Technology Based on Watershed Segmentation and Texture Analysis

Abstract:

Images of glass product that are captured under poor light usually make defected regions covered by dark pixels because of refraction and scattering brought about by defects on glass. On the other hand, image texture makes it unlikely to effectively identify defects accompanied by some texture by using commonly-used image segmentation and defects extraction algorithms. Watershed segmentation algorithm is proposed in this paper to extract catchment basin where defects characteristics in each image region will be calculated with the twofold application of gray level co-occurrence matrix(GLCM) and parameters characteristics . Defects will be finally identified using shape operator.

• Periodical: Applied Mechanics and Materials (Volumes 236-237)
• Pages: 603-608
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.236-237.603
• Online since: November 2012
• Authors: Zheng Tao Zhu, Gang Yang, Bo Zhang
• Keywords: Glass Bottle, Grey Level Co-Occurrence Matrix, Shape Operator, Watershed Segmentation

References

[1] Wang Huarong, Hang Daoping, Xie Haizhi. Watershed segmentation based on morphology applied in glass bottleneck. Computer in Fujian, 98 in 6th session, (2008).

[2] Chen Yuanyan, Jiang Yijun, Wang Qiang. Boundry characteristics and judgement methods for detecting cracks on glass bottle. Drawings 2001 Supplement.

[3] Ding Ting, Yi Dexin, Ding Xiaodan, Fan Hongda. A fast detection algorithm for glass crack.

[4] Jing Hong. Experimental study of glass bubble recognition for grayscale images based on transmission model. Guilin Institute of Technology in 2008 Journal.

[5] Liu Hao, Liu Chun, Hu Cungang. The application of mixed filters into detecting defects on glass bottle.

[6] Yin Jun. Image detection system for medicinal bottles characteristic parameters. Journal of master degree in Hefei University of Technology, (2008).

[7] Zhang Yan, Liu Chun. Defect detection algorithm based on center positioning of bottle circumference. Computer technology and development 6, (2009).

[8] Fu Neng. Research on controlled bottle detection system based on machine vision. Journal of mater degree in Hefei University of Technology , (2009).