Research on the identification method of key parts of ship target based on contour matching

Abstract

A template matching method based on the contour fitting heading angle is proposed for the problem of identifying key parts of maritime ships. First, unmanned boats are used as hypothetical enemy targets to extract outer contours, build a matching template library, and establish relevant kinematic models. Based on the requirements of timeliness and relativism, the judgment conditions for fitting angles and heading angles are given. A traversal fitting framework is established based on the structural similarity index algorithm, and the target matching template is matched based on precise matching results. A motion space that combines position correlation, electro-optical detection distance, and target pitch decoupling is designed to obtain real heading angles. Finally, based on the relative position information of the template’s key parts, the key parts of the target are matched, and the normalized output matching image is obtained. The experiment shows that this method can achieve recognition instructions in real seaways where key parts of the target cannot be extracted due to the large amount of water mist, and it has advantages in timeliness, accuracy, and applicability compared with other algorithms. This method has strong robustness and provides a reference for the identification of key parts of various types of ship targets.