LAUSR

ABSTRACT Kang, L., 2019. Wave monitoring based on improved convolution neural network. In: Gong, D.; Zhu, H., and Liu, R. (eds.), Selected Topics in Coastal Research: Engineering, Industry, Economy, and Sustainable Development. Journal of Coastal Research, Special Issue No. 94, pp. 186–190. Coconut Creek (Florida), ISSN 0749-0208. As human's marine activities, such as marine aquaculture, ocean shipping and coastal facilities construction, flourished world-widely, it is required to monitor the ocean wave dynamics timely and accurately. In this paper, a convolutional neural network (CNN)method was presented to assist the classification of ocean wave. Then it was compared with other traditional models in terms of accuracy and speed. Firstly, the performance of different CNN models was evaluated and the best one was improved with the idea of network in network. Then the best hyper parameters corresponding to the improved model were acquired. Finally, the performance of improved CNN model was compared with that of traditional models on a test database of 26536 image cases. The rate of correct identification in the proposed model is 96.23%, similarly to that of ResNet-152 (97.85%), and it take the least execution times. Overall, this method could be used for near-real-time wave monitoring by efficient classification of ocean wave images.