济州岛南部海域海底声呐图像分析与声学底质分类

Seabed sonar image analysis and acoustic seabed classification in the south of the Cheju Island

东海北部外陆架靠近济州岛南部海域,是黄海槽向冲绳海槽延伸的部分,属于黑潮分支黄海暖流的通道入口,分布着脊槽相间的海底底形,对其海底声呐图像的处理分析及声学底质分类的分析研究,有助于了解该通道海底底形表层纹理特征及沉积物分布规律。基于在济州岛南部海域获取的多波束声呐数据,应用图像处理技术和方法,对数据进行了处理,获得了海底声呐影像图,并对其表层纹理特征进行了描述和分析;同时,基于多波束反向散射强度数据,结合19组海底地质取样数据,建立研究区海底反向散射强度与沉积物粒度特征之间的统计关系模型,并以改进的学习向量量化神经网络方法,实现对海底粉砂质砂、黏土质砂以及砂-粉砂-黏土3种底质类型的快速自动分类识别。

The selected area in this paper is located in the south of the Cheju Island. The study area is part of the Yellow Sea Trough extends to the Okinawa Trough and it＇s in the pathway of the Yellow Warm Current which is one of the branches of the Kuroshio Warm Current. Its seabed sonar image processing analysis and acoustic seabed classification research, contribute to a comprehensive understanding of the characteristics of the channel seabed bedforms surface texture and sediment distribution pattern. With the highprecision multibeam echo sounder sonar da- ta in the research area, we apply image processing techniques and methods and get the seabed sonar image, then we can do quantitative description and analysis to the seabed surface texture features. Based on multibeam echo sounder backscatter strength data and 19 geological seabed sediment sample data, a statistical model which presents the relationship between seabed backscatter signal and sediment type is set up. Using improved Learning Vector Quantization neural network methods,a fast and accurate automatic identification for three seabed sediment types （TS, YS, STY） implementation is feasible.