Egomotion Estimation of an Autonomous Underwater Vehicle Using Sonar Image Sequences

• Main Authors: Chen, Sung-Yueh, 陳松嶽
• Other Authors: J.Guo
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/26018593683726859483

碩士 === 國立臺灣大學 === 造船工程學系 === 86 === Rich environmental information from sonar images can provide the function of navigation and obstacle avoidance for an autonomous underwater vehicle. However, besides noise, reverberation and multipath effects, the images do not describe the location and the shape of objects in water directly. Therefore, a method of sonar image processing in determining the center, major and minor axis and the orientation of objects is desired.The goal of this thesis is to extract the features of objects in images and to calculate the relative motion between the vehicle and targets. After constructing the relations between objects in sonar image sequences, we can calculate their displacements. Because of the noise in images and the approximation of the targets, there are errors in the results of calculation. Finally, we use Kalman filter to estimate the trajectory of the vehicle. The results show that it is feasible to estimate the motion of the vehicle based upon environmental features in image sequences.