为提高舰船监测质量，获取更多的监测信息，进行舰船三维图像重构是必要的。基于该背景，设计视觉传达技术的舰船三维图像自动重构系统。在结合B/S 三层架构的基础上，设计系统框架结构，包括前端硬件采集层，后台软件运行层，二者之间通过无线通信系统连接，在前者完成数据采集之后传输到后者。利用激光雷达采集舰船点云数据，利用CCD相机采集舰船纹理数据，完成舰船视觉图像采集传输单元设计。基于视觉传达设计后台软件处理单元，包括点云数据处理子程序、关键点云提取子程序、点云配准子程序以及三维视觉传达模型构建子程序。结果表明：本文系统应用下，重构的舰船三维图像信噪比达到最大值(26.32 dB)，X，Y，Z 三个方向上的平均误差值达到最小值(0.021 m、0.018 m、0.017 m)，由此说明本文系统重构的三维图像更加清晰，准确度更高。
In order to improve the ship monitoring quality and obtain more monitoring information, it is necessary to reconstruct the ship 3D image. Based on this background, an automatic reconstruction system of ship 3D image using visual communication technology is designed. On the basis of combining the B/S three-tier architecture, the system framework is designed, including the front-end hardware acquisition layer and the background software operation layer. The two are connected through the wireless communication system. After the former completes the data acquisition, it is transmitted to the latter. The laser radar is used to collect the ship point cloud data, and the CCD camera is used to collect the ship texture data to complete the design of the ship visual image acquisition and transmission unit. The background software processing unit is designed based on visual communication, including point cloud data processing subprogram, key point cloud extraction subprogram, point cloud registration subprogram and 3D visual communication model building subprogram. The results show that the designed system should be as follows: the signal to noise ratio of the reconstructed ship 3D image reaches the maximum (26.32 dB), and the average error value in X, Y, Z directions reaches the minimum (0.021 m, 0.018 m, 0.017 m), which shows that the reconstructed 3D image of the designed system is clearer and more accurate.
2022,44(20): 161-164 收稿日期：2022-04-09
DOI：10.3404/j.issn.1672-7649.2022.20.033
分类号：TP25.43
作者简介：高阳(1986-)，女 ，硕士，讲师，研究方向为虚拟现实技术、三维图像、视觉传达技术等
参考文献：
[1] 史玥婷, 任仕伟, 王晓华. 具有噪声鲁棒性的三维磁性纳米粒子成像快速重构方法[J]. 北京理工大学学报, 2022, 42(5): 543–550
[2] 李玉卓, 李新荣, 冯文倩, 等. 面向服装行业的人体三维重构建模研究现状[J]. 毛纺科技, 2022, 50(2): 116–123
[3] 王彩云, 李阳雨, 李晓飞, 等. 基于稀疏性深度学习的航拍图像超分辨重构[J]. 系统工程与电子技术, 2021, 43(8): 2045–2050
[4] 丁玉祥, 卞维新, 接标, 等. 融合邻域回归和稀疏表示的图像超分辨率重构[J]. 计算机工程与应用, 2021, 57(2): 230–236
[5] 刘爱林, 张敬金, 李思宁, 等. 条纹管激光成像雷达目标三维重构快速算法[J]. 中国激光, 2020, 47(1): 289–296
[6] 李宜鹏, 解永春. 基于点云位姿平均的非合作目标三维重构[J]. 空间控制技术与应用, 2020, 46(1): 1–7
[7] 马乐, 陆威, 姜鹏, 等. 基于匹配滤波的Gm-APD激光雷达三维重构算法研究[J]. 红外与激光工程, 2020, 49(2): 153–160
[8] 覃亚丽, 梅济才, 任宏亮, 等. 基于高斯平滑压缩感知分数阶全变分算法的图像重构[J]. 电子与信息学报, 2021, 43(7): 2105–2112
