A performance analysis for real-time flood monitoring using image-based processing

Abstract

Nowadays, various image-based methods have been used in the area of monitoring. Whereas the precision of detection objects and real-time processing are the key issues for many applications. Considering the limitation of the working environment, the higher correctness and faster operating time can guarantee the work efficiency. In this paper, the image-based methods have been studied to monitoring the state of the flood in the real-time system. The performance of each image processing technique has been evaluated based on accuracy and processing time. In the flood monitoring system, the variation of important parameters can cause the change of performance and the effect of the variable parameters has been demonstrated from the experiment results. After comparing to the other image-based techniques, canny edge detection presents the best one, which also has better repeatability with the source image from different locations. Consequently, the improved canny edge detection method has been proved that can work very well on the real hardware in the outdoor environment.