当前位置：首页 > 过刊浏览->2023年45卷7期
基于视频信息的港口滞留船舶检测研究
Research on detection of stranded ships in port based on video information
- DOI:
- 作者:
- 倪慧洋
NI Hui-yang
- 作者单位:
- 江苏航运职业技术学院, 江苏 南通 226010
Jiangsu Shipping College, Management Information Department, Nantong 226010, China
- 关键词:
- 视频信息;港口滞留;船舶检测;形态学滤波;SSD算法;默认框
video information; port detention; ship inspection; morphological filtering; SSD algorithm; default box
- 摘要:
- 本文研究基于视频信息的港口滞留船舶检测方法，通过港口滞留船舶的精准检测提升港口管理水平。利用多结构形态学滤波方法，滤波处理港口视频图像。选取局部自适应阈值分割方法，将滤波处理后的港口视频图像，划分为前景图像与背景图像。将港口视频的前景图像作为SSD算法的输入，SSD算法利用卷积层提取图像特征，生成默认框，利用固定匹配策略，匹配真实框与默认框，将匹配结果传送至预测网络，利用预测网络输出港口滞留船舶检测结果。实验结果表明，该方法有效检测港口视频信息中的滞留船舶，阴天、黑夜等复杂环境下仍然可以精准检测船舶目标。
The detection method of stranded ships in port based on video information is studied to improve the level of port management through accurate detection of stranded ships in port. The port video image is processed by multi - structure morphological filtering method. A local adaptive threshold segmentation method is selected to divide the filtered port video image into foreground image and background image. The foreground image of the port video is taken as the input of the SSD algorithm. The SSD algorithm extracts the image features by using the convolutional layer, generates the default box, matches the real box and the default box by using the fixed matching strategy, sends the matching results to the prediction network, and uses the prediction network to output the detection results of the stranded ships in the port. Experimental results show that the proposed method can effectively detect stranded ships in port video information, and can accurately detect ship targets in complex environments such as cloudy day and dark night.
2023,45(7): 186-189 收稿日期：2022-11-08
DOI：10.3404/j.issn.1672-7649.2023.07.037
分类号：TP391
作者简介：倪慧洋(1990-),男,硕士,讲师,研究方向为港口及航运管理等
