Author(s): Ichiro Fujita; Yuichi Notoya; Mitsuru Shimono
Linked Author(s): Ichiro FUJITA
Keywords: UAV; River Flow Measurement; Image Analysis; STIV; SIFT
Abstract: This paper proposes a new image analysis method to measure river surface velocity distributions from video images obtained from an unmanned aerial vehicle (UAV) .The method is mainly comprised of two procedures, i. e. an image stabilization procedure and a procedure for velocity measurement. In the image stabilization procedure, the blur included in the video images were corrected both accurately and efficiently by combining two novel image processing schemes: the Rotation Invariant Phase Only Correlation (RIPOC) and the Scale Invariant Feature Transform (SIFT) .The surface flow measurements were conducted by applying the space-time image velocimetry (STIV) developed by the authors, which traces the advection of surface features such as ripples or floating objects. The method was applied to a 2014 snowmelt flood of the Uono River in Japan with success of measuring water surface velocity distributions that covers a river reach of about two hundred meters with a width of about 150 meters. The accuracy of the method was evaluated through a comparison with the data obtained by a boat-mounted ADCP that measured the same area in a zigzag manner. Surface velocity distributions measured by STIV using video images taken from a river bank were also used for comparison. The results showed good agreement with the above two methods.