Water surface level profile estimation by image analysis with varying overhead camera posture angle

Abstract

A digital CCD camera was used to capture and record the positions of the water surface at the sidewall of a wave flume by various angles of overhead posture. The wave image data were obtained using spatial transformation, image reconstruction and image detection processes. Two wave detection algorithms, which were the threshold method (TM) and the hyperbolic tangent function (HTF), were used to detect the water surface, and the detected results of the wave were compared with the measurement by a wave gauge. The results show that the wave surface detected by the HTF algorithm is better than that detected by the TM algorithm. The wave periods analyzed by the image data agreed very well with those of the wave gauges and the effect of variation of the overhead posture angle (θ) on wave period measurement was small. Since the pixel resolution increases with the overhead posture angle, however, the mean error of wave height measurement, which was analyzed by the TM algorithm, increased from 1.5% to 2.7% as θ changed from 15° to 45°. When θ was increased to 60°, the mean error of wave height rose to 4.7%. The mean error of wave height measurement, which was analyzed by the HTF algorithm, increased as the overhead posture angle increased, but the deviation was lower than that of the TM algorithm.