The Research On Noise Removal Of Hyperspectral Remote Sensing Data Specreum Domain Based On Wavelet Transform

At present, hyperspectral remote sensing data has been widely used in many fields as a successful remote sensing technology.As a prominent characteristics compared with multispectral remote sensing data, hyperspectral remote sensing data can get high spectral resolution of continuous images,and fully reflect the detail spectra of different surface objects. Under the influence of the instrument and imaging environment,noise is existed both in spatial domain and spectral domain,it’s a serious constraint to hyperspectral remote sensing data who with high spectral resolution for advantage.so,the spectrum domain noise is need to be filter out before hypersoetral data are applied.Having FISS data(one kind of field imaging spectrometer system data)and AVIRIS data(airborne visible infrared imaging spectrometer data)as the research object,the removal of noise in spectral domain and the effect of different noise filter methods on hyperspectral data are discussed. The dissertation can be summarized as following:1Image quality and noise in hyperpsectral data is analysed using standard variance,correlation coefficient and image spectra;2Wavelets achieve good property in both time and frequency resolution,and analyse signals in multiresolution.Therefore,they are widely applied in image processing.Wavelet analysis combined with denosing and compression technology for hyperspectral remote sensing image is studied in this thesis. First of all,according to the selection of wavelet-basis function for hyperspectral image spectrum multi-level decomposition, then on the basis of suited threshold value to shrinking or removal the decomposition wavelet coefficients, finally complete inverse wavelet transform to get the spectral curve without noise.3In order to analysis and evaluation the method in the paper, Select three of the methods who can be applied in spectrum domain noise removal of hyperspectral image. For example,minimum noise fraction (MNF), Saviztky-Golaysmoothing filtering, average filtering. Compare these three methods and the method in the paper in Denoising Effect, spectral comparability and SNR(Signal to Noise Ratio). Draw a conclusion that the method in this paper has achieved a satisfying result and superior to the methods that frequently used in Hyperspectral remote sensing data spectrum domain noise removal.