수중 스테이션 도킹 시 실시간 타원 검색을 이용한 자세 인식 알고리즘

When automatically docking with AUV(Autonomous Under-Water Vehicle) and station, position recognition is an important issue for AUV pose control. The ellipse is good mark for station docking, because it has simple shape information of relative coordinates provide enough for between AUV and station. However, most ellipse detection algorithms have critical issues about complexity and processing time. Today, through the innovation of processors, we can decrease processing time with GPU computing. In this paper, we use Hough transformation for ellipse detection. In the first instance, we get the edge of ellipse candidates. Then we can find the center coordinates of the ellipse described based on the property of ellipse by using three points voting with the Randomized Hough Transformation (RHT). Finally, an analytical solution of the other three parameters of the ellipse is given by coordinate transformation. The RHT is a typical complex and parallel programmable algorithm that can use the GPU (Graphic Processing Unit) computing. Based on this solution, we can compose the real-time robust ellipse detection algorithm for under water station docking.