Busy Fishing Area Recognition based on Improved K-means with Random Walking Centroid.

To overcome the sensitivity of the initial value generated by the K-means algorithm, the selection method of an initial clustering center is improved in this paper. The clustering algorithm of a random walking centroid is used to improve the efficiency, stability, randomness, and diversity of the iterative process and prevents convergence to the loacl optimal solution. The identification model of busy water in fishing areas is developed and applied to the Minnan fishery. The proposed model can accurately identify the to overcome the sensitivity of the initial value generated by the spatial distribution and scale of busy fishing areas. It can also adjust the algorithm parameters according to the merchant ship scale, and generate the targeted recommended route. The results can help maritime security departments. Furthermore, the identification method helps novel safety supervision of the maritime department. [ABSTRACT FROM AUTHOR]