An Image Processing Method for Floating Benzene Slick Detection on Water Surface

Dejun Zhang,Hui Huang,Yichun Shentu,He Wei,Hangzhou Wang,Shuyue Zhan,Hong Song,Jing Xu,Zhongzhi Yao,Hanqing Ji,Caicai Liu,Ren Xu
DOI: https://doi.org/10.1109/oceanse.2017.8084918
2017-01-01
Abstract:In the last 20 years, transport of hazardous chemicals by ship has increased along with the high risk of spill events, which could cause acute and long-term environmental effects. Among hazardous chemicals, benzene and its similar chemicals are carcinogenic. In this paper, the aim of this work is to propose an accurate, quantitative, and non-polluted method to detect floating benzene on a water surface. Benzene of 14 to 18 ml was dropped on the water, and their images were obtained at spectral band of 600nm. Otsu thresholding segmentation was adopted to divide the image into the background and the target briefly, after which morphological operations were used to eliminate the noise components of the background. Then Roberts edge operator was used to detect the edge of the target, and least-squares fitting of ellipses was carried out to link the incomplete edge in order to recover the full region of floating benzene in the image. As a result, the recovered region coincided closely with the area of floating benzene in the origin images. The spreading area of the floating benzene was calculated from the recovered region, and was correlated to the actual amount of benzene. High correlationship was observed, which indicated that the proposed image detection method has potential for floating benzene slick on water surface under mild windy condition.
What problem does this paper attempt to address?