Rapid Calculation Research on Water Area Extraction from ASAR Image

Abstract

Flooded area is a piece of the most important information we’ve got to know in flood supervision and disaster evaluation. And water area extraction is one of the decisive preconditions in confirming the size of flooded areas. With the appearance of remote sensing technology, people could extract the water areas from the space, and the radar images gotten from the flooded area provide us the exact size of water areas. However, the speed of water area extraction affects flood evaluation directly, which matters a lot to the rescue work after flood. To shorten the extracting time of water areas, parallelization process is used as one of the most efficient ways. Based on some normal methods frequently used in water area extraction, such as threshold method, NDVI, and so on, a new way is put forward, that is, to use Self-organized Feature Map to extract the water areas from ASAR images automatically and accurately, and then to analyze the SOM calculation flows to find out what influences the extraction speed, finally to optimize the calculation by using parallel I/O of parallel file system and the asynchronous parallel model of I/O hidden strategy. All of the above enhances the calculation speed so as to ensure that the subsequent flood evaluation and rescue work can be carried out as soon as possible.