Optimizing Image-Based Techniques for River Monitoring: Insights into Graphical Enhancement and Parameter Sensitivity

Abstract

Image-based techniques have gained significant attention for monitoring natural and artificial rivers, thanks to their many advantages over traditional methods. These non-intrusive and highlyversatile optical approaches provide accurate flow discharge measurements, even in challenging conditions like flood events, while ensuring the safety of both operators and equipment. However, the accuracy of optical measurements is affected by several factors, including environmental conditions, river flow characteristics, field acquisition protocols, and the parameterization of the processing software. Image-based techniques follow a three-phase workflow: (i) seeding, (ii) recording, and (iii) processing. Seeding introduces natural or artificial tracers onto the water surface to detect motion. Recording captures video sequences from stationary or mobile platforms (e.g., UASs – Unmanned Aerial Systems). Processing extracts the surface velocity field and flow metrics. The latter phase is divided into three sub-steps: pre-processing, surface velocity evaluation, and postprocessing. Pre-processing includes stabilization, orthorectification, and graphical enhancement; surface velocity evaluation uses correlation-based or similar algorithms to track tracers across frames; finally, post-processing refines velocity data by filtering noise, interpolating missing data, and extracting relevant metrics. Among the steps of optical techniques, graphical enhancement is particularly critical. By increasing the contrast between tracers and the background, it enhances the ability of software algorithms to accurately track motion, thereby reducing errors. However, an inadequate parametric setup of the processing software can also result in the estimation of biased velocities. To investigate these interdependencies, this study conducted a comprehensive sensitivity analysis, evaluating the combined effects of graphical enhancement techniques and processing parameters on the performance of image-based analyses. The analysis compares traditional algorithms with more innovative approaches, including colorspace transformation, and assesses the impact of varying processing parameters under different operational conditions. A dataset of videos acquired from UAS platforms and fixed stations during discharge measurement campaigns on Sicilian rivers, in Italy, was used. The videos were analyzed using PIVlab and SSIMS-Flow software, and the results were benchmarked against ADCP measurements.