A particle-image based wave profile measurement technique

Wave profile measurements are important for computing wave characteristics and for studying the aqueous boundary layer formed beneath surface waves. The measurement technique presented here made use of digital imagery and a detection algorithm referred to as the variable threshold method. The technique can measure wind generated waves as short as 10 pixels (1.44 mm) in wavelength. The average r.m.s. quantization error was found to be +/- 0.29 pixels (+/- 0.04 mm) using simulated wave profiles and the average bias error was estimated to be 0.07 pixels (0.01 mm) from real still water profiles. The magnitude of all other types of random errors was estimated to be approximately +/- 0.64 pixels (+/- 0.09 mm) using real wind wave profiles. A series of morphological operations, used to correct for non-uniform seed densities, improved the accuracy of the detected wave profiles by a factor of five. The variable threshold method detected real wind wave profiles 3.5 times more accurately than the standard constant threshold method and had total r.m.s. errors that ranged from +/- 0.7 (+/- 0.1 mm) to +/- 1.1 (+/- 0.16 mm) pixels.