Study On Edge Extraction Method Of Submarine Geomorphology Unit

The submarine geomorphology provides a basis for studying the movement of the continental plate and the occurrence of natural disasters on the submarine,and has a guiding significance for the installation and laying of offshore platforms and submarine pipelines.So the identification of submarine geomorphology unit and its edge is a fundamental issue.Edge extraction of submarine geomorphology unit is usually execute by observation and experience.This non-automatic identification method is time consuming,laborious,subjective,and inaccurate for most cases.In order to overcome these problems,this thesis studied automatic edge extraction algorithms.The main research content includes:First,this thesis analyzes features of submarine geomorphic unit based on meshed data of MBS.Features of submarine geomorphic unit are calculated including slope,slope of slope,profile curvature and horizontal curvature.Edge and non-edge samples of submarine geomorphology unit are analyzed and compared,Analysis shows that profile curvature of edge section is significant greater than that of non-edge section.Therefore,definition of submarine geomorphic unit edge is proposed based on profile curvature.Then,this thesis proposed an automatic edge extraction algorithm based on region growing rules.Analysis shows that range of edge section is statistically much less than that of non-edge section,and edge of submarine geomorphic unit is defined using profile curvature.So,profile curvature is adopted as an index to separate the edge zone from the whole area.Initial seed points are estimated based on local minimum of profile curvature distribution histogram of the whole areas.For different submarine geomorphic units,non-edge points in the initial seed points are removed according to the variance value of the seed points' neighborhood and the improved K-means clustering.The remaining seed points respond to strong edge zone,and the weak edge zone is estimated by extending the strong edge zone according to a criterion of region growing.The region growing algorithms about single connected and central connected are discussed.Additionally,closing operation is further introduced to supplement discontinuous edge zone.The advantages and disadvantages of two algorithms in searching edge of submarine geomorphic unit are compared and analyzed.Finally,this thesis proposed an improved ant colony algorithm to extract edge of submarine topography unit automatically.The traditional ant colony algorithm has some deficiencies in extracting edge of submarine topography unit,such as long operation time,low accuracy and etc.This thesis improves the ant colony algorithm.Because the seed points in the region growing located near the edge of submarine geomorphic unit,the initial ants are placed at the seed point positions.The initial pheromone matrix is determined by profile curvature values of the neighborhood of the seed points.This algorithm uses local update and global update rules in the process of pheromone updating.According to final pheromone matrix,the threshold is automatically determined for separating the edge zone from the whole area.Edge refinement is exacted for the wider edges.Optimal combination of parameters is discussed in experiments.The verification indicates that the proposed algorithms locate the main edge of the submarine geomorphology unit accurately and efficiently.