摘要
国产HY-1C卫星CZI数据单幅图像具有幅宽、重访周期短、数据处理快等特点。为了提高利用国产卫星数据的利用率,设计了基于HY-1C卫星CZI数据的小目标舰船检测实验方案。首先,对原始卫星数据图像增强,裁剪图像构建数据集;其次使用云检测和海陆分离降低数据处理量;最后,通过YOLOv4算法检测裁剪后图像中的舰船目标,并在整幅卫星图像中显示检测结果。结果表明,HY-1C卫星CZI数据可有效应用到舰船目标检测中,且具有较好的检测效果,拓展了国产卫星数据的应用场景。
The single image in CZI data of the domestic HY-1C satellite has the characteristics of wide width, short revisiting period and fast data processing. In order to improve the utilization of domestic satellite data, a small target ship detection experiment scheme based on HY-1C satellite CZI data is designed. First, the original satellite data image is enhanced and cropped to construct a dataset. Secondly, cloud detection and sea-land separation are used to reduce the amount of data processing. Finally, the ship target in the cropped image is detected by the YOLOv4 algorithm, and the detection result is displayed in the entire satellite image. The results show that the CZI data of HY-1C satellite can be effectively applied to ship target detection, and has good detection effect, which expands the application scenario of domestic satellite data.
作者
任鹏
孙威
吕新荣
REN Peng;SUN Wei;LYU Xinrong(College of Oceanography and Space Informatics,China University of Petroleum(East China),Qingdao 266580,China)
出处
《实验技术与管理》
CAS
北大核心
2022年第11期34-39,共6页
Experimental Technology and Management
基金
山东省自然科学基金项目(ZR2019JQ21)。