Research on Dynamic K-Means Clustering Algorithm in Cyanobacteria Blooms Detection

Abstract:

Cyanobacteria blooms are constantly observed in the coastal waters and pose an enormous threat to public health, economy and ecological environment. The characteristics of blue algal bloom images and feature extraction procedures are analyzed in this paper. The pixel value of Cyanobacteria blooms color images has a significant difference from normal coastal waters images, particularly those of Hue and Saturation. A new method is proposed for Cyanobacteria blooms detection using dynamic K-means algorithm. Experimental results demonstrate the excellent practicability of the proposed detection method. Based on the pixel statistics, it can achieve a highly successful probability of detecting bloom images. Therefore, the proposed detection method can be expected to classify and detect Cyanobacteria blooms in monitoring and forecasting systems.

• Periodical: Applied Mechanics and Materials (Volumes 157-158)
• Pages: 428-432
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.157-158.428
• Online since: February 2012
• Authors: Xu Yang, Xiao Rong Chen, Yi Ting Li
• Keywords: Cyanobacteria Blooms, Image Detection, K-Means Algorithm, Pixel Value

References

[1] Sheng Wang, Nan Jiang, Bin Hu, The remote sensor system of cyanobacteria blooms in TaiHu Based on IDL, Remote Sensor Application, pp.59-63, Feb. (2010).

[2] H.W. Paerl, R.S. Fulton, P.H. MoisanderandJ. Dyble, Harmful freshwater algalblooms, with an emphasis on cyanobacteria, Science. vol. 1, pp.76-113, (2001).

DOI: 10.1100/tsw.2001.16

[3] T. Kutser, L. Metsamaa, N. Strombeck, and E. Vahtmae, Monitoring cyanobacterial blooms by satellite remote sensing, Estuarine, Coastal and Shelf Science, vol. 67, no. 1-2, pp.303-312, (2006).

DOI: 10.1016/j.ecss.2005.11.024

[4] Rafael C. Gonzales, Richard E. Woods, in: Digital Image Process Second Edition (Publishing House of Electronics Industry 2007).