多波束声呐水体影像沉船自动提取方法

Automatic extraction of wreck based on multibeam water column image

为解决人工浏览水体数据耗时且无法保证结果可重复性的问题,提高数据的利用率和定量分析能力,通过分析水体数据的成像机理,提出一种基于单帧水体影像自动提取沉船目标的算法,利用噪声抑制、数学形态学方法检测目标边缘,根据海底统计特性识别海底,再结合深度特性实现海底与沉船目标分离。使用实测数据验证了沉船自动提取算法,且抗噪性较强。通过水体影像提取的目标信息能够弥补水深点目标模型分辨率不足的缺点,为后续目标三维精细建模奠定基础。

To solve the problems of time consumption and inability to guarantee the repeatability of the results when browsing water column data manually and to further improve the data utilization and quantitative analysis capability, we propose an algorithm for the automatic extraction of wreck based on a single water column image (WCI) by analyzing the imaging mechanism of water column data. After detecting the edge of the object by noise suppression and mathematical morphology, the seafloor, which needs to be removed from WCIs with depth, can be identified according to the statistical characteristics. The efficiency of the method has been assessed with the measured data, and its noise immunity is strong. The object information extracted from the WCI can compensate for the drawback of the insufficient resolution of the depth model and lay the foundation for the detailed three-dimensional modeling of the object.