基于SAR图像的自动船只检测模型与算法实现

船只检测是实现船只航行安全的重要方法之一,利用SAR图像可实现船只检测。本文论述了海上船只检测的方法、过程及相应的算法。根据实际检测多位于存在陆地和岛屿的近海图像,利用分形维作为海陆分割的阈值,采用改进的分裂合并算法进行分割,并利用轮廓跟踪及种子填充消除分割遗留的孤立区域,使得海陆分割达到了较好的效果。在比较、试验的基础上,船只检测中采用了改进的CFAR算法,依据视数的不同选择不同的杂波模型,既消除了杂波模型参数计算复杂性,又取得了较好的时效性。通过试验证明,本文提出的算法在实际检测中取得了较好的效果,识别率高、虚景率低且实时性好。

Automatic Ship Detection Model and Algorithm Based on SAR Image

Automatic ship detection is one of the important methods that make ship sailing safely. SAR image can be well applied to detect the ship. The paper introduces the flows, processes and algorithms of ship detection. Because of offing images having lots of islands and land, an improved Split-and-Merge Image segmentation is applied to segment between the land and sea, which uses fractal dimension as threshold. Outline tracking and seed filling are used to clear up the isolated region after segmentation. The algorithms make segmentation between land and sea well. By comparing and analyzing the algorithms of ship detection, the improved CFAR algorithm is applied to ship detection, which is not only robust but also efficient. It chooses the different clutter models according to different looks. The experiment shows that the automatic ship detection method has a high detection ratio and low false alarm.