The Research On Water Information Extraction Method Based On Multisource Remote Sensing Images

With the rapid development of information science and technology in the field of aeronautics and astronautics, the remote sensing technology is more and more important in daily life. At present, the target detection and identification technology with aerospace remote sensing image has been researched by many domestic and foreign experts and scholars in the field.In this paper, based on SAR(Synthetic Aperture Radar) image of China HJ-1C satellite and multi-spectral remote sensing image of China GF-1 satellite water body detection methods are studied:(1) In this paper, in order to exactly extract water body from complex scenery SAR image, firstly, a filter is chosen to reduce the coherent spot noise. And a Sobel gradient image and a contrast stretching transformation image are fused together to enhance image contrast. Secondly, an automatic multi-threshold segmentation based on bimodal iteration method in this paper, which can overcome the shortage that traditional OTSU threshold segmentation algorithm is not suitable for unimodal histogram image segmentation. Then,in order to reduce the false alarm probability, this paper puts forward a kind of eliminating false alarm algorithm based on region topology descriptor.Research results show that the algorithm can effectively eliminate the false alarm of the shadow and has good adaptability and high detection accuracy,which indicates this method can provide some useful land information for detection and assessment of the water.(2) In order to extract water body and make use of the spectrum information of GF-1multi-spectral images,this paper proposes a kind of multi-band relation method based on LBV transform method of GF-1 multi-spectral images. Firstly, this paper calculates the LBV transform formulas of GF-1 satellite multi-spectral images. Secondly, by analyzing typical ground features in L, B and V components, rediation value of water body in B image is greater than in V image and L image.According to this the relationship,water body information can be extracted exactly. The experimental results show that this method can eliminate the shadow and road on the influence of water body information extraction. And the proposed method can accurately extract water of large area and small area, such as rivers, lakes and ponds and so on. This method is simple and has the strong adaptability.