Underwater terrain navigability analysis based on image processing and fuzzy decision
SONG Zi-qi;BIAN Hong-yu;Adam Zielinski;ZHANG Zhi-gang;Science and Technology on Underwater Acoustic Laboratory, Harbin Engineering University;College of Underwater Acoustic Engineering, Harbin Engineering University;Department of Electrical and Computer Engineering, University of Victoria;
Underwater terrain-aided navigation method based on multi-beam bathymetric measurements and image-matching techniques requires navigability analysis of the region where an autonomous underwater vehicle will be applied. Image feature extraction is necessary to calculate the distinctiveness of this region, and comprehensive fuzzy evaluation can then be implemented using these features. By defining the histogram complexity, the feature of the seabed fluctuation is obtained from the gray scale histogram of a terrain image. Root-mean-square contrast and Laplacian(8-neighborhood differential) operator sum reflect the underwater topographic relief in global and local respectively. Two-dimension entropy of terrain images provides the information quantity of the corresponding geography. Since the navigability rank of the possible working areas calculated from a single image feature might be inconsistent with the one from the other, the fuzzy decision is used to obtain comprehensive results. By using the terrain matching method based on texture features of underwater terrain images, the simulation analysis is made from actual multi-beam bathymetric data, which shows that the local region ranking first in the chosen terrain image can promise a positioning error of less than 2 m, verifying the effectiveness of the proposed navigability analysis method.