基于深度学习的复杂气象条件下海上船舶识别

为提高复杂气象条件下海上船舶的识别效果,本文通过暗通道先验去雾算法减少云雾遮挡对目标识别的影响,使用基于深度学习的YOLO改进算法提高目标识别效果。结果表明：本文采用的算法在中国航天科工四院指挥自动化中心的模拟海事数据集上,4类船舶目标识别的mAP (Mean Average Precision)达到89.98%,超过了对比的其他目标识别算法;针对数据集中的云雾遮挡图像,暗通道去雾处理后,目标识别的mAP从53.25%提升到69.35%。可见本文提出的算法可以满足复杂气象条件下的海上船舶识别的需求。

Recognition of Marine Vessels under Complex Weather Conditions based on Deep Learning

In order to improve the recognition effect of marine ships under complex meteorological conditions, dark channel defogging algorithm is used to reduce the influence of cloud cover on target recognition , and the improved YOLO algorithm based on deep learning was used to improve the recognition effect. The results show that in the simulated maritime data set of the command automation center of China aerospace science and technology institute, the Mean Average Precision of the four types of ships" target recognition is 89.98%, which is higher than the comparison of other target recognition algorithms. For cloud and fog shielding images in data sets, the Mean Average Precision of target recognition was increased from 53.25% to 69.35% after dark channel defogging. It is concluded that the algorithm proposed in this paper can meet the requirements of marine vessel identification under complex meteorological conditions.