通过对无人艇航行轨迹虚拟重构，提高对无人艇的目标跟踪调度能力，提出基于多准则粒子滤波下的无人艇航行轨迹虚拟重构方法。采用融合双模态特征检测方法实现对无人艇航行轨迹的图像检测，结合视觉分割和分块特征匹配实现对无人艇航行轨迹重构过程中的显著性检测，采用多准则粒子滤波方法进行寻优迭代，根据深度学习和粒子滤波增强结果，实现无人艇航行轨迹虚拟重构和模态跟踪识别。测试表明，采用该方法进行无人艇航行轨迹虚拟重构的视觉增强能力较好，期望值和实际值拟合性能较好。
By virtual reconstruction of the navigation trajectory of unmanned boats, the target tracking and scheduling ability of unmanned boats is improved. A multi criteria particle filter based virtual reconstruction method for unmanned boat navigation trajectory is proposed. The fusion bimodal feature detection method is used to achieve image detection of unmanned boat navigation trajectory. Visual segmentation and block feature matching are combined to achieve saliency detection during the reconstruction process of unmanned boat navigation trajectory. Multi criterion particle filter method is used for optimization iteration. Based on the results of deep learning and particle filter enhancement, virtual reconstruction and modal tracking recognition of unmanned boat navigation trajectory are achieved. Tests have shown that the visual enhancement ability of using this method for virtual reconstruction of unmanned boat navigation trajectories is good, and the output peak signal-to-noise ratio is high.
2023,45(9): 94-97 收稿日期：2022-12-02
DOI：10.3404/j.issn.1672-7649.2023.09.020
分类号：TP391
基金项目：吉林省教育厅科学技术研究资助项目（2021JB505L10）
作者简介：李克玲(1979-),女,硕士,讲师,研究方向为计算机及应用
