Abstract
Fluorescent and colored sands have been used to trace sediment movement in coastal field observation. A huge cost is required to detect these tracers from sediment samples by use of conventionally human inspection. Many computer-assisted digital image analysis techniques for the detection system have been developed in the context of price reduction of digital imaging devices in recent years. However, these methods give insufficient accuracy of detection. For instance, it would be difficult to find one tracer particle from the large-volume sediment sample by using fully automated image analysis. In this study, we developed a detection method of colored sand tracer based on a combination procedure with computer image analysis and human visual inspection. The digital-image-scanning procedure was determined experimentally by laboratory tests with regard to sediment sample weight, means of pasting sediment and its image resolution. Furthermore, final check of results from the image analysis made high accuracy of detection by the human inspection with trimmed-image outputs from the analysis program. This method was well balanced for cost-benefit among analysis time, the accuracy and human workload. In particular, it is efficient for colored sand detection tasks with a large number of sediment samples because of untired computer-assisted procedure.
