A Lightweight Ship Detection Method in Optical Remote Sensing Image under Cloud Interference
Jinxiang Yu, Xiyuan Peng, Shaoli Li, Yibo Lu, Wenjia Ma · 2021
Ship detection in optical remote sensing image is faced with challenges of high detection false alarm caused by cloud interference, and the contradiction between detection accuracy and computation workload. In this paper, a lightweight anti-cloud ship detection method is proposed. The framework of subgraph classification and mapping reduces the computation workload, and the classifier based on joint feature of gray level co-occurrence matrix, local binary pattern and support vector machine achieves a low detection false alarm. Compared with YOLOv3, SSD and MobileNet-SSD methods, experimental results show that the proposed method outperforms in terms of false alarm rate and computation workload.