This paper focuses on the manipulation of flexible objects by a dual-armed robot. The objective of this research is to detect adjacent rope in a tabletop knotting operation and to operate the dual-armed robot to resolve the adjacent state. First, the shape of a rope on the tabletop is obtained from a thin line image. Then, the rope thickness is measured from the coordinates of the obtained rope shape, and the adjacent points are detected. Finally, based on the obtained data, the system performs a de-adjacency operation by holding the rope with one hand and inserting it into the adjacent point with the finger of the other hand.