基于热红外视频图像监测的海面溢油识别技术研究

针对现有海面溢油检测技术难以在石油泄漏初期（尚未形成海面大规模油膜覆盖）及时发现油膜的难题，本文在前期基于热红外图像测算海面油膜面积方法研究的基础上，结合油泄漏至海面后油膜的扩散特征，提出了一种基于热红外视频图像监测油膜面积变化以及时识别海面溢油的方法。首先，基于单帧热红外图像处理算法提取海面前景区域（包含油膜区域与相似物干扰区域）并计算各区域所代表的实际物理面积。基于视频图像处理技术跟踪测算前景区域中各连通区域的实际物理面积变化情况，根据各连通区域的面积变化率识别前景区域中是否存在油膜，从而判断海面是否发生溢油。实验结果表明:所提出的方法能有效识别不同黏度的石油泄漏至海面形成的扩散油膜，在水面包含波浪与相似物干扰时也具有良好的识别精度。该方法适用于特定场景下（如码头、船舶等）的溢油事故的鉴别，能为溢油事故的及时发现和预警提供技术支持。

Research on identification of marine oil spill based on thermal infrared video image monitoring

When oil spill occurs and the large scale covering on the sea surface has not been formed, it is hard to find oil film by existing oil spill detection technology. To solve this problem, a novel method for discriminating of oil spill by monitoring the area of oil film is presented based on thermal infrared video image, which combined with the diffusion characteristic of oil spill. Firstly, foreground regions (real oil film region and look-alikes interference region) on the sea surface are extracted and the actual physical area of each region is calculated based on single-frame thermal infrared image processing (i.e., the pixel area calculation method from the previous research). According to the video image processing, the change of the actual physical area of each region is tracked in real-time. The area change rate threshold is set to discriminate whether oil film on the foreground regions, then whether oil spill happened can be determined. The experimental results show that the proposed method can effectively discriminate oil film formed by different viscosity of oil and maintain good identification accuracy under sea surface with waves and floating objects. This strategy is suitable for specific scenes such as wharves and ships, and also can provide technical support for pollution control of oil spill.