Attitude estimation from noisy sonar images

Several researches estimate locations and attitudes of underwater vehicles through seafloor patterns in onboard video or sonar images. However, seafloor maps are needed in advance and the images are usually noisy and obscure. In practical images only objects to be identified are the sea surface and the seafloor. In this paper, a new attitude estimation technique based on acoustical image processing is presented. It is a three dimensional expansion of Hough transformation which is robust for noisy images. After the procedure explanation, numerical simulation conditions and results are shown. An elementary optimization trial is also presented.