基于特征匹配的舰载对陆导弹目标识别模型

针对传统基于前视模板的匹配算法中难以直接识别与跟踪建筑等目标的问题, 提出基于特征匹配的对陆导弹目标识别模型。该模型通过对末制导导引头图像预处理, 利用改进YOLOv3深度学习目标检测算法和改进Deeplabv3+深度学习语义分割算法来识别目标区和烟雾区, 采用并行法排除烟雾遮挡对目标识别的干扰, 最终判别分析规则判断模型是否识别成功。仿真实验结果表明，该模型能够快速有效精确地完成对陆地目标的识别, 兼具较好的抗烟雾干扰能力, 有利于提高对陆导弹的目标识别水平与作战效果。

Target recognition model of ship-to-land missile based on feature matching

Aiming at the problem that traditional forward-looking template matching algorithm is difficult to identify and track targets such as buildings directly, a target recognition model for land missiles based on feature matching is proposed. Through prepocessing of images of terminal guidance seeker, the model takes advantage of the improved YOLOv3 deep learning target detection algorithm and Deeplabv3+deep learning semantic segmentation algorithm to recognize the target area and the smoke area. Parallel method is used to eliminate the interference of smoke occlusion on the target recognition. Finally, the discriminant analysis rule is used to judge whether the model is successfully identified. The simulation experiment results show that the model can recognize the land target quickly, effectively and accurately, which has good anti-smoke interference ability. It helps to improving the level of target recognition and combat effectiveness of the land missile.