Ship target classification based on Hu invariant moments and ART for maritime video surveillance

In the intelligent video surveillance system used at sea, the target classification is the basis of the video content analysis and the target's behavior identification. Thus, this research work is about to classify the ship targets at sea by using shape features. This paper extracted the regional shape eigenvalues of ship targets by Hu invariant moments and ART coefficients, selected 7 regional shape eigenvalues of highest distinction from the 43 regional shape eigenvalues extracted to constitute ship shape feature vector. On this part, this paper establishes the shape feature library for different types of ships (such as bulk carrier, container ship, tanker, etc.), proposing a classification method for the ship targets based on KNN template by matching and conducting the verification. It is proved that the 7 ship regional shapes eigenvalues can effectively distinguish different types of ship targets. Besides, the ship classification's accuracy rate of the method proposed by this paper is very high.