It is of great significance to seabed exploration for the development and utilization of marine resources,marine engi-neering construction,and national defense security.As an acoustic device capable of surveying the distribution of bottom sedi-ment in the shallow surface of the seabed,the accuracy of bottom sediment identification currently depends on the subjectivity of the operator,with poor reliability.In order to improve efficiency and interpretation accuracy,it is necessary to further study the intelligent identification model for the bottom layer boundary.In the paper,an improved region growth algorithm suitable for seabed bottom layer boundary recognition without human intervention is proposed.That is,based on the study of grayscale mapping and noise elimination,the skeleton information of the image layer boundary is extracted using an iterative maximum class difference algorithm,and then the skeleton information is used as an initial growth point and the growth direction is cor-rected using rheological properties.At the same time,the algorithm combines grayscale weighted mapping curves and peak valley wavelength constrained growth neighborhoods to segment layer boundaries,extract edges,and connect them into lines,thereby seabed bottom layer boundary recognition is ultimately achieved.The experimental results of shallow section survey data of Lianyungang port waterway show that the improved region growth algorithm can effectively identify the boundary of the bottom layer,and its recognition accuracy reaches centimeter level,which can meet the requirements of seabed sediment inter-pretation and analysis.
关键词
浅地层剖面仪/底质层界/最大类间差算法/区域生长算法/自动识别
Key words
sub-bottom profiler/bottom layer boundary/maximum class difference algorithm/regional growth algorithm/au-tomatic identification
维普
万方数据