研究基于改进卷积神经网络的舰船实时目标跟踪识别技术，满足复杂背景下舰船目标跟踪识别的高精度需求。利用时空上下文算法确定舰船图像中，舰船目标与周围区域的时空对应关系，依据对应关系构建舰船图像目标置信图，将置信图中具有最大似然概率的区域，作为舰船目标的初定位区域；利用卷积神经网络搜索初定位区域，通过卷积层和下采样层的运算，识别舰船目标的精确位置；依据舰船目标精确位置识别结果，选取相关滤波算法，设置相关图中最大响应值位置作为舰船目标最新位置，输出舰船目标实时跟踪结果。实验结果表明，该技术在云雾遮挡、弱光照等复杂背景下，均可以精准跟踪识别舰船目标，舰船目标跟踪识别的平均覆盖率高于95%。
The ship real-time target tracking and recognition technology based on improved convolutional neural network is studied to meet the high precision requirements of ship target tracking and recognition under complex background. The spatio-temporal correspondence between the ship target and the surrounding area in the ship image was determined by using the spatio-temporal context algorithm. According to the correspondence relationship, the ship image target confidence map was constructed, and the region with maximum likelihood probability in the confidence map was set as the initial location region of the ship target. The convolutional neural network was used to search the initial location area, and the precise location of the ship target was identified by the operation of the convolutional layer and the downsampling layer. According to the accurate position recognition result of ship target, the correlation filtering algorithm is selected, and the position of the maximum response value in the correlation graph is set as the latest position of ship target, and the real-time tracking result of ship target is output. Experimental results show that the technology can accurately track and recognize ship targets under complex background such as cloud occlusion and weak illumination, and the average coverage of ship target tracking and recognition is higher than 95%.
2022,44(21): 152-155 收稿日期：2022-06-22
DOI：10.3404/j.issn.1672-7649.2022.21.031
分类号：TP391
作者简介：于国莉(1977-),女,硕士,副教授,研究方向为人工智能及移动应用开发
参考文献：
[1] 刘坤, 于晟焘. 基于卷积神经网络的云雾遮挡舰船目标识别[J]. 控制与决策, 2021, 36(3): 661–668
LIU Kun, YU Shengtao. Obscured ship target recognition based on convolutional neural network[J]. Control and Decision, 2021, 36(3): 661–668
[2] 郭晨, 简涛, 徐从安, 等. 基于深度多尺度一维卷积神经网络的雷达舰船目标识别[J]. 电子与信息学报, 2019, 41(6): 1302–1309
GUO Chen, JIAN Tao, XU Congan, et al. Radar HRRP target recognition based on deep multi-scale 1D convolutional neural network[J]. Journal of Electronics & Information Technology, 2019, 41(6): 1302–1309
[3] 马剑飞, 颜冰, 林春生, 等. 基于多通道卷积神经网络的磁性舰船目标运动参数估计[J]. 国防科技大学学报, 2020, 42(4): 78–84
MA Jianfei, YAN Bing, LIN Chunsheng, et al. Motion parameter estimation of magnetic ship target based on multi-channel convolutional neural network[J]. Journal of National University of Defense Technology, 2020, 42(4): 78–84
[4] 张云, 化青龙, 姜义成, 等. 基于混合型复数域卷积神经网络的三维转动舰船目标识别[J]. 电子学报, 2022, 50(5): 1042–1049
ZHANG Yun, HUA Qinglong, JIANG Yicheng, et al. Recognition of 3D rotating ship based on mix-CV-CNN[J]. Acta Electronica Sinica, 2022, 50(5): 1042–1049
[5] 付哲泉, 李尚生, 李相平, 等. 基于高效可扩展改进残差结构神经网络的舰船目标识别技术[J]. 电子与信息学报, 2020, 42(12): 3005–3012
FU Zhequan, LI Shangsheng, LI Xiangping, et al. Ship target recognition based on highly efficient scalable improved residual structure neural network[J]. Journal of Electronics & Information Technology, 2020, 42(12): 3005–3012
[6] 尹雅, 黄海, 张志祥. 基于光学遥感图像的舰船目标检测技术研究[J]. 计算机科学, 2019, 46(3): 82–87
YIN Ya, HUANG Hai, ZHANG Zhixiang. Research on ship detection technology based on optical remote sensing image[J]. Computer Science, 2019, 46(3): 82–87
[7] 王俐云, 方峰, 伊成俊, 等. GEO遥感卫星在舰船目标检测跟踪中的应用分析[J]. 航天器工程, 2021, 30(5): 133–139
WANG Liyun, FANG Feng, YI Chengjun, et al. Application research of GEO remote sensing satellite in ship detection and tracking[J]. Spacecraft Engineering, 2021, 30(5): 133–139
