Rapid Flood Detection Using Remote Sensing Images

Abstract

The main motive of a flood detection system is to detect and forecast flood to alert the public from natural calamities and threats. Although many flood detection systems are already available but there is a need of a system with high computational and processing power with reduced cost. The main purpose of this paper to implement high performance flood detection system using Graphics Processing Unit(GPU). Graphics processing units have very high computation power as it comes with hundreds of cores in contrast to CPU. Also they are less expensive and can be used as a default computing environment for parallel processing. Flood detection is a very critical task requiring instant and robust processing. The remote imagery obtained from satellite is processed to detect the extent of flood. The parallel computation is exploited to process large amount of pixels at a time. Algorithms are implemented both on CPU as well as GPU to evaluate the performance in terms of speed.