Water level detector for early warning systems using color difference measurement

The rise in water level of river may lead to changes in geological and artificial environments, thus triggering the possibility of different scales of geohazards on the slopes adjacent to the river. Therefore, initiation to detect and monitor the water levels of river with remote sensing data should be developed. In this research, a method in detecting the rise in river water level was developed. The principle of this method was to compare the camera’s image keypoint features from before the overflow of river with the occurrence of the overflow of river in a determined span of time. This research used Color Differences to compare the images. The ΔE* distance metric was used to calculate the change in structural information of the images. It will be used as a reference in determining the river water level during normal conditions, rise and overflow. Keypoint value of an image may change if geohazard on the riverbank undergo a change caused by the inundation over the grass, soil, rocks on the riverbank. Determination of river water level was successfully done with the category of image difference consisting of normal level (Diff ≤ 20%), stand-by level (20% < Diff ≤ 30%), warning level (30% < Diff ≤ 40%) and alert level (Diff ≥ 50%). This system only works well in bright lighting conditions and not in dim conditions.