Discrimination Of Algal-Blooms Using Spaceborne SAR Observations Of Great Lakes In China

In recent years,with the economic development,human activities in the lake basin have intensified.The spectral characteristics of algae are similar to those of plants and are quite different from those of water,therefore optical remote sensing can effectively identify water bloom.However,optical remote sensing can only be observed in the daytime,and is restricted by cloud,rain and other meteorological factors.Synthetic aperture radar(SAR)has the characteristics of all-weather,all-day,weather free and penetrating,which makes up for the deficiency of optical remote sensing.With the development of SAR technology,more and more attention has been paid to the monitoring of water environment by microwave remote sensing.Therefore,the study of water bloom monitoring and identification based on SAR is of great significance.This paper takes taihu lake,a typical water body in China,as an example,and chaohu and danjiangkou reservoirs as verification to carry out a study on the detection and identification of water bloom based on SAR images.Dark speckle in SAR image is the main research object."Dark spot" refers to the low-gray area in SAR images caused by the reduction of radar backscatter due to the water bloom.However,targets such as "low wind speed" water surface will also reduce radar backscatter,forming a similar phenomenon of dark spots,causing interference for the identification of water blooms using SAR images alone.In this paper,integrated application of image segmentation algorithms related to spots of preliminary segmentation,using synchronous optical image tag has split the properties of the spots,according to the scattering characteristics of the surface regions in SAR image,geometry feature,texture feature,such as multiple factors analysis,determine the characteristic parameters of can maximize recognition blooms,finally construct the SVM based SAR image recognition model of water China,experiments show that the SAR can be used as identification of lake water blooms,and have the potential to improve the recognition accuracy.The main work and innovation of this paper can be summarized in the following three aspects:1.For SAR images with low water China wind area target is difficult to determine type of problem,put forward a kind of quasi synchronous optical and SAR image contrast pretreatment method,can more accurately determine the type of spots of attributes.2.Proposed a SAR image segmentation algorithms spots,based on k-means clustering segmentation algorithm,combined with the statistical characteristics of image gray level Multi-threshold Otsu algorithm,the improved k-means K value clustering center is difficult to determine and choose difficult problem.At the same time in the segmentation process combined with the waterland segmentation enhancement algorithm universality.Finally,combined with the regional growth algorithm and morphological filtering method of self-determination center,the precise segmentation results of dark spots are obtained.3.On the basis of c-svm,a water bloom classification and identification model was established.First of all,the identification and classification basis of the two types of dark spots were determined by analyzing the dark spots and similar dark spots.Subsequently,the model was trained by cross validation with the corresponding features of the sample as parameters,and the identification of water bloom was completed after parameter optimization.The applicability of the model has been verified by experiments on the water bloom in danjiangkou and chaohu waters.