The Application Of Unmixed Pixels With Constrained Least Squares Algorithm In Hyperspectral Remote Sensing

Remote sensing technology is developed to be a comprehensive modern physics,space technology, computer technology, mathematics and geography of detectingtechnology of the new study in the1960s. the applications of hyper-spectral remotesensing in geology survey is one of the most successful aspect.generally, mineralspectrum absorption peaks in30nm,hyper-spectral data of spectral resolution is wideband of remote sensing data10times, in the wide band can’t response spectrumcharacteristic, but in hyper-spectral image is easy to identify, which fundamentallychange the optical remote sensing images from the extraction of geologicalinformation quantity and quality. Spectral resolution is just to identify it with hyperspectral.no matter in information quantity or quality. it changed the situationcompletely in remote sensing.De xing tailings Located in Shangnao city of Jiangxi province, the copper mine isthe richest one of Asian. mineral is big, centralized reserves, burial shallow, strippingratio small, the tailings of copper mine has useful elements, the value of the tailings ispriceless.The key point of this thesis is to introduce tailings data after using high-passfilltering, low pass filtering and after principal components extraction and MinimumNoise Fraction, and then use the data of Pure Pixel index, and by using least squaremethod, without restraint, abundance sum-to-one constraint, abundance nonnegativityconstraint method,and then use these methods to decomposition of pixel in image,Wechoose four miner such as Arsenopyrite Hs62.3B、Azurite WS316、ClinochloreGDS159、Bronzite HS9.3B,we get abundance figure and a comparison error of threedifferent methods After mixing solution, the non-negative constrained least squaresalgorithm is the best one of the three methods. And the condition of abundancesum-to-one constraint is difficult to realize. At last,we realize the three algorithm withIDL language, and show the figure of the abundance,in contrast,we can safely come toa conclusion that the algorithm of non-negative constrained least squares is the bestone.