육상에서의 환경 구성을 통한 광학 카메라 및 소나 센서 모델 기반 고신뢰도 수중 소나 이미지 생성 기법

Abstract

One challenge in developing an underwater sonar algorithm is the lack of data. This paper proposes a method for efficiently synthesizing arbitrary underwater sonar image. The proposed method reconstructs a model in three dimensions from an optical image using neural network, and then generates sonar images according to sonar sensor’s imaging model. As a result, by physically realizing the specific condition on laboratory, underwater sonar data set can be efficiently constructed.