Sediment Classification Based On Acoustic Methods

The classification and identification of the seabed geological type is the basic of ocean scientific research, it has significant meaning to classify and identify the marine sediments exactly and quickly in military and location. Taking the acoustic method , through distant detection of the character of the marine sediments to learn its physical quality has the characteristic of efficiency ,economical , getting data continuously, which combined with a certain traditional method and optical observation offers a quickly and reliable way in the classification of the marine sediments.For the complicated trait of the seabed, certain method and certain characteristic quantity can only identify some part of the marine sediments. It needs several characteristic quantities, even some different methods to availably identify marine sediments. so exploring new and more effective ways to classify the characteristic quantities has further meaning .In the thesis, several classification methods based on statistic characteristics are presented. At the domain of time, frequency and frequency-time, the classification of ocean bottom is analyzed respectively. In time domain, the characteristic parameters correlated to duration of echo signal are resorted with the methods of power curve and phase plane. In frequency domain, sub-band power is used in resorting to the character of frequency alternation. In the domain of time and frequency, short time FFT of received signals, continuous wavelet transform, discrete wavelet transform are introduced to. Several transforms are made to hydrophone receiving signals, such as, with singular value decomposition, the resident is also classified. In order to determine the efficiency of the presented methods, in some criticisms are introduced. Comparison of every method shows that discrete wavelet transform combined with SVD has the best efficiency of classification. Based on the computer simulation of the above methods, the experimental data are used to do the research of resident classifications. The final results meet with the bottom sampling well.