視界情報による潜水艇の制御情報取得に関する考察 : 第2報

概要

論文の詳細を見る
In the previous study, basic methods to estimate motion or attitude of the vehicle (camera),i.e. ego motion, from video image data obtained continuously by a video recorder are investigated, and several alternative methods to estimate the vehicle motion are evaluated by sample pictures. From the study, possibility of the ego motion estimation at the sea bottom, where few landmarks exist and images are not so clear, is assured for the case human operator can recoguize the motion visually, using full information and selecting appropriate estimating points. In this study, more effective iterative estimating algorithm based on the Extended Kalman Filter is studied, and the availability of the method is evaluated by applying them to continuous video image data obtained from a camera moving in a two dimensional horizontal plane. By choosing appropriate filtering pamameters, it is possible to get the accurate estimation within l/3 of computing time needed for the previous methods.
社団法人日本船舶海洋工学会の論文