Intelligent algorithm of warship’s vital parts detection, trajectory prediction and pose estimation

Accurate detection and attack of warship’s vital parts can effectively improve the damage efficiency of anti-ship missile. Aiming at the problems of low detection accuracy on vital parts and insufficient accuracy of guidance error, this paper proposes an algorithm of warship’s vital parts detection, trajectory prediction and pose estimation based on deep learning. The deep semantic information and shallow positioning information are integrated, and the SoftPool modules are used to preserve fine-grained features. The detection accuracy of multi-angle and multi-scale warship’s vital parts is improved. Combining the detection results with the warship’s space structure can establish the mapping relationship, which is used to calculate the three-dimensional position and posture of the seeker. The long short term memory network is introduced to mine the space-time characteristics of key-points to realize the dynamic trajectory prediction on multi-scale warship. Experimental results show that this algorithm has high accuracy in detection of warship’s vital parts and trajectory prediction. The posture estimation results of the seeker are precise. The situational awareness requirement in complex marine battlefield of autonomous self-piloted anti-ship missiles is satisfied in independent penetration perspective. © 2023 Beijing University of Aeronautics and Astronautics (BUAA). All rights reserved.