Identification And Extraction Of Typical Vegetation On Qinghai-Tibetan Plateau Based On Spectral Matching Technique

Spectral matching technology is the key technology of target recognition of imaging spectrometry.According to different applications,how to choose different spectral matching technology and whether to develop better technology has been a research hotspot.Restricted by the special geographical environment of the Qinghai Tibet Plateau,at present,there are few researches on vegetation classification in Qinghai Tibet Plateau.But the vegetation types and their changes have important significance for the study of the evolution of the Tibetan plateau ecological environment and its response to global change.In this paper,HJ-1A hyperspectral images and ASD measured hyperspectral data were used to identify and extract the vegetation in the typical area of Hoh Xil plateau in Qinghai Tibet Plateau.At first,the spectral data of the measured hyperspectral data are smoothed and denoised by using the Savitzky-Golay filter,and the problem of the vertical stripes of HJ-1A hyperspectral data is solved by using the Globally Eliminate Strip method.And then this paper establishes the Plateau Vegetation spectral database based on the measured spectra after pretreatment.Finally,this paper uses the Cross Correlogram Spectral Matching,Spectral Angle Matching,Spectral Feature Fitting,Wavelet Transform Spectral Matching and other spectral matching methods to match identification test and comparative study on extraction effect for little wormwood,stipa purpurea,carex moorcroftii and other Plateau Vegetation.Results show:(1)The Spectral Characteristic Parameters and Spectral Absorption Index can better extract the unique spectral characteristics of the green and healthy vegetation;Cross Correlogram Spectral Matching need the optimal weights in distinguishing similar spectra to the peak value and the skewness coefficient of the cross correlation spectrum;Wavelet Transform Spectral Matching is limited by the error threshold of spectral matching data,for the complex environment of the Tibetan Plateau and the low spatial resolution of the satellite remote sensing data,the method of Wavelet Transform Spectral Matching display limitations;Spectral Angle Matching and Spectral Feature Fitting are common spectral matching method,Spectral Angle Matching gives more consideration to the shape characteristics of the spectrum,but Spectral Feature Fitting focuses on the nature of the spectrum,and can effectively extract the spectral features of the spectrum with obvious spectral absorption characteristics.(2)In this paper,through the experimental comparison of the above spectral matching methods,it is found that they can better identify the spectrum oflittle wormwood in the research area;Wavelet Transform Spectral Matching and Spectral Feature Fitting can better identify the spectrum of stipa purpurea in the research area,Cross Correlation Spectrum Matching cannot distinguish between the spectrum of stipa purpurea and carex moorcroftii.At the end of the paper,the vegetation coverage area of little wormwood and stipa purpurea was extracted by Spectral Angle Mapping method,and the results were validated by 1:100 million Chinese Vegetation Map.