Utilization of new echo attributes for acoustic seabed classification and characterization

单波束回声测深仪是一种标准的海洋测量工具，可产生用于绘制海底水深图的回波。这些回波还包含有关海床的信息，既来自从海床表面散射回来的能量，也来自紧邻海床下方的沉淀物体积。在有限的情况下，通过对这些回波的直接估计来表征沉积物的性质。然而，声学海底分类系统使用的一种更实用的方法是从海底回波中提取称为特征的描述性统计信息。这种方法在当今的商业系统中很常见。海底分类系统将分割算法应用于特征集，以将海底划分为具有相似声学特征的分段。通过地面真实信息，这些分段往往成功地与海床类型类别相关联。该项目的目标是利用基于理论的回波属性来改进海床声波分类，并通过直接估计海床属性来研究海床特征。这些属性中最有希望的是回声持续时间，但其他属性，如相对回声幅度/能量的测量，集合之间或单个回波内的可变性测量，以及频谱特征，如功率谱矩，将被检查。这些属性可能与海底属性有关，例如体积衰减、海底粗糙度和声阻抗。这项研究将使用广泛接受的海底后向散射模型Boris产生的模拟数据和Patricia Bay的地面现场数据。

The single-beam echosounder is a standard marine surveying tool that generates echo returns used to map seabed bathymetry. These echo returns also contain information about the seabed, both from energy scattered back from the seabed surface and from the sediment volume immediately below the seabed. In limited situations, sediment properties have been characterized by direct estimates from these echoes. However, a more practical approach utilized by acoustic seabed classification systems is to extract descriptive statistics known as features from seabed echoes. This approach is common to present day commercial systems. Seabed classification systems apply segmentation algorithms to sets of features to divide the seabed into segments of similar acoustic character. Through ground truth information, these segments are often successfully related to classes of seabed types.The goals of this project are to use theoretically based echo attributes to improve acoustic seabed classification, and to investigate sea bed characterization by estimating sea bed properties directly. The most promising of these attributes is echo duration, but others such as measures of relative echo amplitude/energy, measures of variability between sets or within single echoes and spectral characteristics such as moments of the power spectrum will be examined. These attributes may be related to seabed properties such as volume attenuation, seabed roughness, and acoustic impedance. Both simulated data generated with the widely accepted seabed backscatter model BORIS and groundtruthed field data from Patricia Bay will be used in this study.